Hormone Markers

Testosterone

Total testosterone

The total amount of testosterone in the blood, both the fraction bound to carrier proteins and the small free fraction. In men about 95% of it comes from the testes, with the remainder made from adrenal precursors. In women the sources are split completely differently. Roughly 25% comes from the ovaries, roughly 25% from the adrenal glands, and about 50% from peripheral conversion of circulating precursors into testosterone in fat, skin and liver.

Tested when investigating low energy, low libido, poor recovery or fertility, and to monitor testosterone replacement therapy. Persistently low testosterone with symptoms is called hypogonadism, and it can originate in the testes or in the pituitary, which is why LH and FSH are read alongside it. Levels above the normal range in someone not on treatment are uncommon and need a cause found. In someone on treatment it is a finding their prescriber would want to know about.

What testosterone actually does, in both men and women, and why both ends of the range cause trouble, is covered in Sex hormones, in the Body Guide. This entry is about reading the number.

Timing is not optional. Testosterone follows a daily rhythm, highest in the morning and falling through the day. How far it falls depends on age. In men in their 30s the drop between 8am and 4pm is roughly 20% to 25%. By 70 it is only about 10%, because the rhythm flattens as men get older. An afternoon result that comes back clearly normal is therefore reassuring. An afternoon result that comes back low cannot be acted on and needs repeating in the morning, because men sampled through the day routinely produce an afternoon reading in the hypogonadal range while every one of their morning readings is normal.

Food matters too. A glucose drink has been shown to drop testosterone by about a quarter, and testosterone is still suppressed 2 hours later. A normal meal seems to do the same thing. This is similar to the drop you see in testosterone from morning peak to the afternoon. But the guidelines split on how hard to push fasting. European and American guidance wants a fasted sample. The British Society for Sexual Medicine says patients should ideally fast but calls insistence on it unreasonable, citing a UK lab review that found no significant fasting effect. On the hour they agree more closely. The British Society for Sexual Medicine asks for the sample between 7am and 11am, the Society for Endocrinology narrows that to between 8am and 10am, fasted and rested, and European urology guidance moved in 2026 from a 7am to 11am window down to 7am to 10am.

Where the numbers come from

In men. Two British guidelines set the lines, and they do not agree with each other.

The British Society for Sexual Medicine, in its 2023 guideline, puts the treatment line at 12 nmol/L (346 ng/dL). A man with symptoms below that, or with a free testosterone below 0.225 nmol/L (225 pmol/L), usually warrants treatment. Above 12 nmol/L he does not. Where total testosterone lands between 8 and 12 nmol/L (231 to 346 ng/dL), it asks for free testosterone to be checked before deciding. Between 8 and 14 nmol/L (231 to 404 ng/dL) it allows a 6-month trial of treatment on symptoms.

The Society for Endocrinology, in a 2022 guideline, works to lower figures. It takes them from the European Male Ageing Study, where sexual symptoms became more likely below a total testosterone of 8 nmol/L (231 ng/dL) whatever the free testosterone, or below 11 nmol/L (317 ng/dL) where calculated free testosterone was also below 220 pmol/L. It presents those as symptom thresholds and not as a definition, asks for characteristic clinical features alongside the biochemistry, and warns that the higher the threshold is set the greater the risk of a wrong diagnosis.

Abroad the split runs the same way. The American Urological Association uses a single figure, 300 ng/dL (10.4 nmol/L). The Endocrine Society declines to set one at all, asking instead for symptoms together with a testosterone that is unequivocally and consistently low, and publishing a reference range in healthy young men whose lower limit is 264 ng/dL (9.2 nmol/L). The European Association of Urology and the Fifth International Consultation on Sexual Medicine have both landed on 12 nmol/L, the same figure as the British Society for Sexual Medicine.

So a man with symptoms at 10 nmol/L is a treatment candidate under one British guideline and not under the other, and nothing on his report will say which one his clinician follows. Every one of those lines was drawn to decide who gets treated, not to describe what anyone should aim for. No professional body publishes an optimal testosterone level, and nothing here is quoted as one.

A man at the top of the range and a man near the bottom are in visibly different physiological positions, and that difference is real. What the evidence does not support is that moving higher within the range buys anything measurable. Excess mortality appears only below about 7.4 nmol/L, and excess cardiovascular mortality only below about 5.3 nmol/L. Cancer death follows the same shape, rising below about 8.6 nmol/L. Above those points the line flattens out, so the risk sits at the bottom of the range and does not run as a slope across it.

The reverse case usually gets left out, so it is stated here too. Testosterone's reputation for danger runs well ahead of its record. The largest randomised trial of testosterone treatment, in men who already had cardiovascular disease or a high risk of it, found no excess of heart attacks, strokes or cardiovascular deaths, and no excess of prostate cancer among men screened before entry. It did find more clinical fractures, though not more major osteoporotic fractures, along with more atrial fibrillation, more acute kidney injury and more pulmonary embolism in the treated men. On symptoms it improved sexual activity, sexual desire and hypogonadal symptoms overall, and it did not improve erectile function scores or prevent diabetes. Treatment comes out safer than the caution around it suggests, and narrower in its benefits than the marketing around it suggests.

A history of prostate cancer is listed as a reason not to treat in most of the guidelines above, and in 2026 that was put to a controlled trial for the first time. Men who had had surgery for low-grade or favourable intermediate-grade disease, with an undetectable PSA for at least 2 years and a low testosterone, were treated for 12 weeks. Sexual activity, desire, body composition and fitness improved, erectile function did not, and there was no biochemical recurrence in either group. The trial was short and it was small, and the people who ran it say plainly it settles neither recurrence nor long-term safety. It is still the first evidence of any kind on a question that had otherwise been answered by assumption.

Two cautions are real. Genetic studies point towards higher lifelong testosterone raising coronary risk through blood pressure, and in February 2025 the American regulator dropped its cardiovascular boxed warning from testosterone products while adding a new class-wide warning about raised blood pressure. Levels above the physiological range carry their own consequences. Haematocrit and blood pressure are the two things that need watching.

A repeat sample is standard before acting on a low result, and every guideline named above requires one. Around 3 in 10 men whose first testosterone comes back in the hypogonadal range are normal on the second.

In women, the question reverses. Testosterone is measured to ask whether there is too much, not too little. The Society for Endocrinology's 2025 guideline on evaluating androgen excess in women sets total testosterone above 5 nmol/L (about 144 ng/dL) as the level warranting further investigation, including confirmation by mass spectrometry and imaging of the ovaries or adrenal glands. That figure decides who gets investigated and is not itself a diagnosis. In postmenopausal women the published literature points to a lower figure, around 2.2 nmol/L (about 63 ng/dL), as the point at which underlying pathology becomes more likely, and that one comes from the literature and not from a guideline.

No professional body publishes a threshold for low testosterone in women, and no diagnostic cut-off exists. The global consensus position states it outright. No blood level of any androgen separates women with sexual dysfunction from women without it. A reader looking for a low-testosterone number for women will not find one anywhere, and anyone quoting one is not quoting a guideline.

That was tested again in 2026, in community-living women with every steroid measured by mass spectrometry and the obvious confounders accounted for. No hormone was associated with sexual desire. The conclusion drawn was that these measurements should not form part of a routine assessment of a woman presenting with sexual concerns.

Treatment is a separate question from measurement. A testosterone cream, AndroFeme 1, received a UK marketing authorisation in July 2025 for one narrow indication, low sexual desire after the menopause. The licence arrived without a diagnostic threshold attached, because there is not one to attach.

Where testosterone is prescribed to a woman, measurement does have a defined job. The global consensus position asks for a level before starting, a repeat 3 to 6 weeks after, and monitoring every 6 months, to confirm the level stays inside the female physiological range. The British Menopause Society says the same, and adds that total testosterone reflects the response better than free testosterone or the free androgen index. That is a monitoring use and it sets no threshold for starting, because there is not one to set. Both also say that where mass spectrometry is not available, a standard immunoassay is adequate for the narrow purpose of ruling out a high level before starting and a supraphysiological one during treatment.

The route changes what the number means

Testosterone reaches a woman's bloodstream by cream, by gel, by injection or by a pellet implanted under the skin, and the 4 do not produce the same picture on a blood test.

A licensed female cream at the standard dose produces an average level of about 1.2 to 1.9 nmol/L, inside the female range. That is the route every professional body endorses, and it has drawbacks that rarely get stated alongside the endorsement. Testosterone on skin transfers to anyone in close contact with that skin, which is why the male gels carry a warning in the US about virilisation in children exposed to a parent's dose. How much is absorbed varies between people and with where it goes on, it comes off with washing and sweat, and it has to be done every day.

A pellet sits at the other extreme. A 100 mg implant peaks around 4 weeks at roughly 3 to 5 times the top of the female range, then falls away over the following months, so a sample at week 4 and a sample at month 5 from the same woman on the same pellet differ by about two and a half times. The people who implant them describe their own peak levels as 4 to 6 times the upper limit of normal. Two women given the same pellet have come back differing tenfold. A pellet also cannot be removed, so a high result has nowhere to go.

Injections are where the official position and the arithmetic come apart. Every body that addresses route says injections are not recommended because they produce levels above the female range. Read closely, that objection is written about the concentration and not about the needle. The cholesterol evidence quoted alongside it says something narrower, because oral testosterone worsened the lipid profile while non-oral routes did not, and the non-oral arms were creams and patches rather than injections. Nobody has published what a female-sized injected dose actually does to a woman's blood level, so the objection rests on pharmacology borrowed from male doses. The practical difficulty is real and it is not physiological. Testosterone for injection is sold at male strength, so a female-sized dose is a fraction of a millilitre and hard to draw accurately.

Outside all of that, a large number of women, mostly in the US, use small weekly or twice-weekly injections under the skin and report it suits them better than a daily cream. No study has measured what that produces, so neither the practice nor the objection to it rests on female data. What can be said without a study is that the height of the peak and the depth of the trough follow from the dose and the gap between doses, so a small dose given often produces a flatter line than a large one given rarely. That is arithmetic rather than a finding.

What it all means for reading a result is simple enough. On a cream, a level can be read against the female range and taken at face value. On a pellet, the answer depends almost entirely on how many weeks it has been. On an injection there is no published female curve to read it against, so the date and size of the last dose matter more than the number.

For women still cycling, the 2025 guideline asks for a morning, fasted sample in the early follicular phase. It also undoes a piece of conventional advice. Measured by mass spectrometry, testosterone itself varies little across the menstrual cycle. Cycle timing matters more for androstenedione and 17-hydroxyprogesterone, which run higher in the luteal phase. The same guideline says the combined oral contraceptive should be stopped 3 months before androgens are measured.

Why the assay matters more here

Testosterone is measured by immunoassay almost everywhere, including ours, and immunoassays are least accurate at low concentrations, which is where female testosterone sits. Two separate things go wrong, and they are the same 2 things that go wrong with oestradiol in men, with the sexes and the hormones swapped. Each assay was built around the concentrations found in the other sex.

The first is cross-reactivity. The antibody is raised against testosterone, but it also binds steroids that are shaped like testosterone and counts them as testosterone. The manufacturer publishes how much each one registers. The largest is 11-beta-hydroxytestosterone, at about 20%, followed by 11-ketotestosterone and androstenedione at roughly 3% each and dihydrotestosterone at about 2%. Norethisterone, a progestogen used in several contraceptives, registers at about 3%. In a man none of this matters much, because his testosterone is several times higher than the steroids being mistaken for it. In a woman that relationship inverts. Androstenedione alone runs at several times her testosterone concentration, so a small percentage of a large number is added to a small one. The steroid people expect to be the problem, DHEA-sulphate, is not one, despite circulating in the micromolar range, because it registers at 0.001%.

The second is the bottom of the measuring range. The manufacturer states a limit of quantitation, the level below which the test can tell you something is there without reliably telling you how much. For testosterone that limit sits above the 5th centile of the manufacturer's own female reference interval. Part of the normal female range is, by the maker's own criteria, not quantifiable.

Mass spectrometry has neither problem, which is why the guideline asks for confirmation by it before an immunoassay result in a woman is treated as final.

Biotin. Testosterone is one of the markers biotin affects, and on this test it reads falsely high. Biotin, at the top of this guide, has the detail.

The androgens nobody measures

There is a second family of androgens, made by the adrenal glands, that almost no panel includes. They are the 11-oxygenated androgens, and the main ones are 11-ketotestosterone and 11-beta-hydroxyandrostenedione. In women they circulate at higher concentrations than testosterone itself, and in the renamed ovarian syndrome they run higher still.

Two things follow. The first is the cross-reactivity above, because these are the steroids the testosterone assay partly mistakes for testosterone. The second is that a normal testosterone does not always mean normal androgens. The rises track the picture a woman presents with, so women whose total testosterone is raised show broad increases in 11-oxygenated androgens, while women with the clinical signs alone and a normal total testosterone have not been found to differ from women without the condition. Nothing on a standard panel measures them either way. In men the family behaves in the opposite direction to the classic androgens as weight rises, going up while testosterone goes down.

No reference intervals exist for any of them, no consumer panel sells them, and we do not either. They are here because they explain part of why the testosterone number behaves as it does.

What a blood test can and cannot establish

Raised androgens are one of the three Rotterdam criteria used to diagnose polyendocrine metabolic ovarian syndrome (PMOS), previously polycystic ovary syndrome (PCOS). The other two are irregular or absent ovulation, and ovarian morphology. Two of the three are required.

That third criterion changed in 2023. In adults, a raised anti-Müllerian hormone is now accepted in place of the ultrasound, and the guideline is specific that one or the other should be done and not both, because doing both drives overdiagnosis. The substitution applies to adults only. In adolescents the ovarian criterion is not used at all.

So blood can now contribute two of the three, androgens and anti-Müllerian hormone standing in for the scan. What blood cannot supply is the cycle history. A raised androgen result on its own still establishes nothing, and two criteria met on paper is a long way short of a diagnosis made by someone who has also taken a history.

The condition was renamed by international consensus in May 2026, with backing from patient and professional organisations including the Endocrine Society and Verity in the UK. There is a 3-year transition period, with the new name fully in place for the 2028 guideline update, which is why both names appear here. The reason for the change matters. The ovarian criterion has always described many small follicles and not cysts, and the consensus documents state there is no increase in abnormal ovarian cysts in the condition. The old name misled on exactly that point.

The rename did not change the diagnostic criteria. Those remain the ones set out in the 2023 international guideline, whose next update is due in 2028.

Units

The UK reports total testosterone in nmol/L and the US in ng/dL. Multiply nmol/L by roughly 28.8. So 10 nmol/L is about 288 ng/dL, 20 nmol/L is about 576 ng/dL, and 1000 ng/dL is about 34.7 nmol/L.

Free testosterone

Calculated free testosterone

The fraction of testosterone not bound to a carrier protein and therefore available to act on tissue. Most testosterone in blood is bound to SHBG or albumin and is not immediately usable.

This figure is calculated and not measured. It is derived from total testosterone, SHBG and albumin using a published equation. That is standard practice worldwide, and the number is only as reliable as the three inputs behind it.

It matters most where SHBG is unusually high or low, because total testosterone can look entirely normal while the usable fraction is not.

Where the numbers come from

We calculate free testosterone from total testosterone, SHBG and albumin using the Vermeulen equation, which is the standard approach in UK practice, and we do so for women as well as men. Calculated free testosterone is not the same thing as directly measured free testosterone. Direct measurement by equilibrium dialysis with mass spectrometry is the reference method and is not commonly available.

That distinction turns out to matter. When free testosterone is measured directly in older men, the measured figure tracks erectile function, intercourse frequency, orgasm frequency and desire. Total testosterone tracked none of the 4. The calculated figure tracked 2. So the calculation is an improvement on total testosterone and it is not a substitute for the measurement.

The British Society for Sexual Medicine uses free testosterone below 0.225 nmol/L (225 pmol/L) as supporting evidence for treatment where total testosterone sits in the 8 to 12 nmol/L zone. The Society for Endocrinology uses 220 pmol/L instead, and uses it differently, as part of a symptom threshold alongside a total testosterone below 11 nmol/L rather than as a supporting check. The 5 pmol/L between the two British guidelines makes no practical difference to anyone. It is a reminder that these lines are drawn by committees and not found in nature. There is no equivalent figure for women.

American reporting uses pg/mL or ng/dL. For free testosterone, 225 pmol/L is about 65 pg/mL, or about 6.5 ng/dL. Those two American units differ from each other by a factor of 10, so any figure found online is meaningless until you know which one it is in.

The combined oral contraceptive

SHBG is one of the three inputs, so anything that changes SHBG changes the calculated result even when testosterone production has not changed at all. The clearest example is the combined oral contraceptive, which raises SHBG substantially and pushes calculated free testosterone down with it. Anyone testing androgens while on the combined pill should know the result is difficult to interpret before they pay for it.

Biotin. This figure is calculated from total testosterone and SHBG. Biotin pushes total testosterone falsely high and SHBG falsely low, so it moves this number up twice over. Biotin, at the top of this guide, has the detail.

Free androgen index

FAI

A ratio, and not a measurement. Total testosterone divided by SHBG, multiplied by 100. Because both figures are in the same units they cancel, so the result has no units at all.

It attempts to answer the same question as free testosterone, which is how much testosterone is actually available to tissue and how much is held by its carrier protein, using 2 numbers instead of 3. Free testosterone is calculated from total testosterone, SHBG and albumin. The free androgen index leaves albumin out.

It was devised for investigating androgen excess in women, and that remains where it is most used.

Anything that moves SHBG moves this number in the opposite direction, whether or not androgen production has changed at all. SHBG falls with obesity, insulin resistance and an underactive thyroid, which pushes the index up. It rises with oestrogen, including the combined oral contraceptive, with an overactive thyroid and with age, which pushes the index down.

Male and female results are not on the same scale. Testosterone runs roughly an order of magnitude higher in men against a broadly similar SHBG, so a male index runs far higher than a female one and the two must never be read against the same range.

Where the numbers come from

No professional body publishes a committed threshold for it. The figures in circulation come from individual lab ranges and commercial testing sites, and not from a society position, and because both inputs are assay-dependent the ratio is more method-dependent than either number alone. The range printed on your own report is the one that applies.

Its validity in men is openly disputed. The UK guidance on androgens in men, published in 2010, describes the free androgen index as of limited value in men, and suggests it might be considered only where a total testosterone is not diagnostic, particularly in obesity. Published work going back 3 decades has argued it was never valid in adult males at all. It was devised for female hirsutism and its extension to men was never formally justified. The work behind the equation we use for calculated free testosterone found the free androgen index was not a reliable estimate of free testosterone, while the calculated figure was.

The failure is predictable and not random. The free androgen index over-estimates free testosterone at low SHBG, and increasingly so the lower SHBG goes. Low SHBG travels with obesity and insulin resistance, so the index is least reliable in the people most likely to be testing. The Society for Endocrinology's 2025 guideline on androgen excess in women puts a number on it. Below an SHBG of 30 nmol/L the index becomes less accurate and may overestimate free androgens, so the guideline asks for caution before using a raised index on its own to support a diagnosis of androgen excess.

So where both appear on the same report, calculated free testosterone is the figure UK practice uses in men, and the one the published thresholds attach to. The same 2025 guideline finds the calculated figure agrees better with the reference method across the whole span of SHBG values, which is where the index falls down. The free androgen index is reported alongside it and reads best as a rough companion number.

Biotin. The same applies here and for the same reason. This number is total testosterone divided by SHBG, biotin pushes the first falsely high and the second falsely low, so it moves up twice over. Biotin, at the top of this guide, has the detail.

Oestradiol

E2

The main oestrogen in both women and men. In women it drives the menstrual cycle and is central to bone, cardiovascular and cognitive health. In men it is made from testosterone by the aromatase enzyme, and it does real work of its own.

In women the number cannot be read without knowing the cycle day. Days 2 to 5 is the standard baseline, and at that point its main job is to validate the FSH result taken alongside it.

In men it is read alongside testosterone, because the balance between them matters more than either alone.

Where aromatase sits in the body, and which effects belong to oestradiol and not to testosterone, is covered in Sex hormones, in the Body Guide. This entry is about the number itself, and the number has problems the Body Guide does not go into.

Both directions cause problems in men. Too high is associated with breast tissue development, fluid retention and mood changes, and no published level attaches to any of those. Too low happens when aromatase is suppressed too aggressively, and that direction has actually been measured. In the trial that separated the two hormones, men pushed below roughly 37 pmol/L (10 pg/mL) lost more than twice as much sexual desire as men at the same testosterone whose oestradiol stayed above it. The bone results from the same work landed on the same figure. A man who crushes his oestradiol while raising his testosterone keeps the muscle and loses the libido, the erections, the fat control and, in time, the bone.

Joint pain and low mood. Men who suppress their oestradiol report both, consistently, and the mechanism is not mysterious. Oestrogen receptors sit in cartilage, in the synovium and throughout the brain in both sexes. The formal evidence comes from women. Joint pain affects roughly half of women taking aromatase inhibitors for breast cancer, it is one of the most common reasons they stop the drug, and it is understood as oestrogen withdrawal acting directly on joint tissue. Nobody has run the equivalent study in men, so what exists is a well-described syndrome in women, a receptor distribution that does not care which sex you are, and a large body of consistent reports from men. That falls short of proof and stands a good deal above anecdote.

Where the numbers come from

No professional body publishes an optimal oestradiol level for men, and none publishes a treatment threshold either. The British Society for Sexual Medicine, the European Association of Urology and the Society for Endocrinology all decline to set one, and the Society for Endocrinology's position statement on confirming male hypogonadism does not mention oestradiol at all. The 37 pmol/L figure above is a trial finding and not a guideline position, and it is quoted here as the former.

The assay problem in men

Oestradiol is measured by immunoassay almost everywhere, including ours, and immunoassays are at their weakest at the concentrations found in men, in children and after the menopause. American lab medicine guidance is blunter about this than anything in UK guidance and says plainly that oestradiol should not be ordered by immunoassay in men, children or postmenopausal women.

The reason is in the manufacturer's own documentation. The current-generation Roche assay, which is a good one and better than the generations before it, detects down to about 18 pmol/L. Its limit of quantitation, the level below which it can tell you something is there without reliably telling you how much, is about 92 pmol/L. The median healthy man sits at about 91 pmol/L. Roughly half of healthy men therefore have an oestradiol the assay cannot reliably quantify, and the 37 pmol/L figure that actually matters clinically sits far below that line.

Cross-reactivity is the other half of it, and here the modern assays have improved. Testosterone and DHEA-sulphate do not register on the current Roche method at all. What still registers are oestrogen metabolites, at low percentages that become less trivial when the oestradiol itself is low. Any antibody-based method carries some of this. Mass spectrometry does not, which is the entire argument for it.

Biotin. Oestradiol is one of the markers biotin affects, and on this test it reads falsely high. Biotin, at the top of this guide, has the detail.

Why we do not offer mass spectrometry

Liquid chromatography-tandem mass spectrometry is the reference method for oestradiol and it is routine in the US. In the UK it runs as an accredited NHS supraregional referral service through a small number of specialist centres, with published reference ranges and a turnaround of around 3 weeks. What does not exist is a consumer route to it.

The logistics are most of the reason. The sample has to be spun, separated and frozen within 24 hours, so it cannot travel in an ordinary postal kit, and one referral costs more at wholesale than an entire 8-marker hormone panel. No UK consumer testing company sells a mass spectrometry oestradiol. Not us, and as far as we can establish, not anyone.

So if you need one, the route is a clinician arranging it through the NHS referral network. That network was built to answer specialist clinical questions and was never built to be bought.

After menopause, oestradiol is low and stable, and every problem above returns for the same reason. There is now a harmonised international reference range for it, recalibrated against the reference measurement procedure so that it does not belong to any one machine. It runs from 4 to 67 pmol/L, with a median around 18, and in women with a body mass index under 30 it runs from 4 to 46. It tracks body mass index and not age.

Put that next to the limit of quantitation above and the problem states itself. The whole of that harmonised postmenopausal range sits below the level at which the assay most labs use can reliably say how much is there. A postmenopausal oestradiol from a routine immunoassay is a number the method was not built to produce.

The guideline that says not to use this test

NICE guideline NG23 on menopause recommends not using oestradiol to identify perimenopause or menopause in anyone aged 45 or over. It is named alongside anti-Müllerian hormone, inhibin A, inhibin B, antral follicle count and ovarian volume. In that age group menopause is identified clinically. Someone 45 or over not using hormonal contraception who has not had a period for at least 12 months has enough for the diagnosis without any blood test, and recently developed vasomotor symptoms alongside a change in the cycle is enough to identify perimenopause.

Units

The UK reports oestradiol in pmol/L and the US in pg/mL. One pg/mL is 3.67 pmol/L. So 30 pg/mL is about 110 pmol/L, and 100 pmol/L is about 27 pg/mL.

Progesterone

A hormone produced after ovulation by the corpus luteum. Its main role outside pregnancy is to prepare and maintain the uterine lining in the second half of the cycle.

It is not only a female hormone. Men make progesterone too, in the adrenal glands and testes, where it is a precursor to testosterone and cortisol. It also acts in the brain, and men's levels are rarely looked at despite that. Low progesterone in women is associated with cycle problems and difficulty maintaining early pregnancy.

The cycle day decides what this number means. Progesterone is low for most of the month and peaks about 7 days after ovulation. Measured at the wrong point it will read low whether or not anything is wrong.

Where the numbers come from

The timing matters more than the number. Current NICE fertility guidance, which replaced the older guideline in March 2026, asks for a sample in the mid-luteal phase and gives day 21 of a 28-day cycle as the example. Day 21 is not a fixed date. It is what the mid-luteal point works out at in a 28-day cycle, and the general rule is 7 days before the period is due. On a 35-day cycle that is day 28, and a sample taken on day 21 will read low for reasons that have nothing to do with ovulation.

NICE does not publish a progesterone number at all. Neither the current guideline nor the one it replaced states a threshold for confirming ovulation. The 30 nmol/L figure used across UK practice comes from a 1982 paper that looked at cycles which resulted in conception, and suggested 30 as a working figure partly to allow for assay variation. It has been repeated ever since. It is a convention, and a reasonable one, and it is not a guideline position and nobody should present it as one.

The banding that appears on commercial testing sites, where above 30 confirms ovulation, 10 to 30 is equivocal and below 10 is anovulatory, is not in any professional body document we could find. It exists because testing companies need something to print in the interpretation box.

Other published figures sit lower, and the reason is that bodies differ on what the test is for and not on the physiology. The American Society for Reproductive Medicine treats a luteal progesterone above roughly 9.5 nmol/L as indicating that ovulation happened. The World Health Organization's evidence review worked to about 19 nmol/L. A figure of 10 ng/mL, about 32 nmol/L, is widely quoted in American practice, though no professional body sets it. The lower figures answer whether she ovulated. The higher ones answer whether the luteal phase was a good one. Those are different questions.

The assay adds its own spread. Progesterone immunoassays disagree with each other between analysers by enough that a 30 nmol/L cut-off is not the same number on every platform, which is a further reason the range on your own report is the one that applies.

Biotin. Progesterone is one of the markers biotin affects, and on this test it reads falsely high. Biotin, at the top of this guide, has the detail.

Progesterone is also released in pulses and not steadily, and the swing across a single day in the luteal phase is very large, so a single value is a snapshot of a fluctuating number. The World Health Organization suggests repeating the measurement where the first one indicates no ovulation, which markedly reduces the chance of calling an ovulatory cycle anovulatory.

The guideline that says not to use this test

The European Society of Human Reproduction and Embryology suggests not routinely measuring mid-luteal progesterone in women with regular menstrual cycles at all, on the basis that a regular cycle already tells you what the test would. The World Health Organization does suggest using it, and grades that as a conditional recommendation on very low certainty evidence. Two international bodies, looking at the same evidence, landing in different places.

What else exists, including things we do not sell

Ovulation can also be confirmed at home by measuring pregnanediol glucuronide, the urinary breakdown product of progesterone. Several monitors do this, some combining it with oestrogen and luteinising hormone across the cycle. One is cleared by the American regulator specifically as an aid to confirming ovulation, and devices of this type are sold in the UK. Urinary pregnanediol glucuronide tracks ovulation closely when it is checked against repeated ultrasound scanning. The caveat is that the fertility tracking market is crowded and inconsistent, with wide variation in what these devices claim to detect and how well they have been tested, and few of them checked against ultrasound at all. If you buy one, buy one that has been.

Salivary progesterone remains a research tool and not a validated clinical test.

Luteal phase deficiency

The idea that a weak luteal phase causes infertility or miscarriage, and can be found with a progesterone test, is widely repeated and not established. The American Society for Reproductive Medicine reviewed it again in 2026 and concluded it has not been proven to be an independent cause of either, and that no diagnostic test for it has been shown to be reliable. Where it is defined at all, it is defined clinically, as a luteal phase lasting 10 days or less, and not by a progesterone number.

In men

No published threshold exists, and there is no established use for a single progesterone measurement in a man who feels well. The science has not been done, as opposed to having been done and come back negative.

Sex Hormone Binding Globulin

SHBG

A protein made in the liver that binds testosterone and oestradiol and carries them around the blood. Hormone bound to SHBG is held in reserve and is not acting on tissue.

It is measured alongside testosterone because it changes how a total testosterone result should be read. High SHBG can leave total testosterone looking normal while very little is available, and low SHBG can do the reverse.

How the binding arrangement works, and why two of the numbers on your report are calculated from it and not measured, is set out in Sex hormones, in the Body Guide.

Where the numbers come from

No professional body publishes a treatment threshold for SHBG. Labs publish reference intervals, and those are assay-dependent, so the range on your own report is the one that applies.

SHBG is a liver readout before it is a hormone carrier, and that is what makes the list of things that move it stop looking arbitrary. Almost all of them act on a single control switch inside the liver cell, and what sets that switch is how much fat the liver is actively manufacturing. Sugar and fructose, and the lipogenesis they drive, turn it down. Thyroid hormone turns it up, which is why an overactive thyroid raises SHBG and an underactive one lowers it. Oestrogen turns it up, including the combined oral contraceptive. Androgens turn it down, testosterone therapy included. Age raises it.

Liver disease moves it in both directions and which one depends on the disease. Cirrhosis raises SHBG. Fatty liver, which is far more common, lowers it. The single line saying liver disease raises SHBG is the version usually printed, and it misleads the larger group.

So what does a low SHBG actually mean? It travels with obesity and insulin resistance, and it predicts the later development of type 2 diabetes. What is unresolved is whether SHBG itself does anything, or whether it is simply reporting on the liver.

The argument that it is mostly reporting is the stronger one. Testosterone treatment pushes SHBG down, sometimes to very low levels, and yet reduces liver fat instead of increasing it. If a low SHBG were driving fatty liver, that should not happen. The genetic work arguing the other way leans on markers that independently affect liver fat, and its own authors say they cannot rule out that this explains their result. No trial has ever raised SHBG deliberately to see whether anything improves.

The shape of the relationship is not a straight line either. Men in the middle of the SHBG range have the lowest risk of dying of cancer, with higher risk at both ends, and a lower SHBG also predicted a later diagnosis of prostate cancer where testosterone itself predicted nothing at all. So a low SHBG is not simply better or simply worse, and which part of the range you sit in matters more than which direction you have moved.

None of which changes what to do about it. An unexplained low SHBG on a hormone panel is a reason to look at waist, liver markers and blood sugar. It is one of the few results on a hormone panel that is mostly telling you about something else.

Because SHBG feeds two other numbers on the same report, a change in SHBG alone moves them both without anything having happened to testosterone production.

Weight loss drugs move this number

Anyone losing a significant amount of weight on a GLP-1 drug is likely to see SHBG rise, and the reason is the switch described above. Losing fat improves insulin sensitivity, less insulin reaches the liver, and the liver makes more SHBG. The drug is not acting on SHBG. In men who are already metabolically healthy these drugs do not change reproductive hormones at all, which is the clearest sign that what moves the numbers is the metabolic repair and not the molecule.

The knock-on effect on a hormone panel catches people out. Total testosterone rises, because fat tissue was converting it to oestradiol and because the same insulin signal was suppressing production. SHBG rises at the same time. Calculated free testosterone is worked out from both, so it often barely moves even though total testosterone has climbed. Someone reading only the total sees a transformation. Someone reading all 3 numbers sees something more modest. LH and FSH hold steady or rise, which is the opposite of what happens on testosterone treatment.

Biotin. SHBG is one of the markers biotin affects, and on this test it reads falsely low. Because it feeds the 2 numbers above, a biotin-affected SHBG drags both of them up with it. Biotin, at the top of this guide, has the detail.

Follicle Stimulating Hormone

FSH

A pituitary hormone that drives egg development in women and sperm production in men.

In women it starts the follicles growing each month. It is measured on days 2 to 5 to assess ovarian reserve and investigate cycle problems, and it rises as ovarian function declines. Persistently high FSH can therefore reflect declining ovarian function, though see below on when it is and is not useful for identifying menopause.

In men it acts on the Sertoli cells in the testes, which nurse developing sperm through to maturity. High FSH in a man with a low sperm count usually means the testes themselves are struggling, and low FSH points at the pituitary instead. European urology guidance adds a caveat that matters to anyone facing surgical sperm retrieval. FSH does not reliably predict whether sperm will be found, and men with a high FSH and no sperm in the ejaculate can still have pockets of active sperm production in the testis.

Where the numbers come from

NICE guideline NG23 on menopause is unusually direct about this, and it runs against how the test is usually sold. It is also old in the part that matters here. NG23 was updated in November 2024, but that update dealt with treatment and not with how menopause is identified, so the recommendations quoted here predate a fair amount of what is on this page. That does not make it wrong. It does mean a guideline is a snapshot of what was settled when it was written, and the literature does not wait for the next revision.

In otherwise healthy people aged 45 or over with menopause-associated symptoms, NICE says to identify perimenopause and menopause without lab tests, on clinical history alone. The physiology behind that is straightforward. FSH swings considerably in the years leading up to menopause, so a single level in that window can read almost anything and would not change what happens next.

NICE says to consider serum follicle stimulating hormone to confirm menopause in only two situations: in people aged 40 to 45 with menopause-associated symptoms including a change in their cycle, and in people under 40 in whom menopause is suspected. It also says not to use it at all to identify menopause in someone using combined oestrogen and progestogen contraception or high-dose progestogen.

Where premature ovarian insufficiency is suspected under 40, NICE requires symptoms together with a raised follicle stimulating hormone on two samples taken 4 to 6 weeks apart. One raised reading is not enough. The European guideline on the same condition attaches an actual number, above 25 IU/L, alongside disordered cycles for at least 4 months, and asks for the repeat only where there is doubt. The two therefore differ on whether the second sample is mandatory, which matters if you are told a single result settles it.

The staging system used in menopause research also uses above 25 IU/L to mark the late transition, and it is careful to call that supportive and not diagnostic, for a reason given in the next section.

Two things that can make this number wrong

Biotin. FSH is one of the markers biotin affects, and on this test it reads falsely low. Biotin, at the top of this guide, has the detail.

The analyser. FSH and LH are calibrated against different international reference preparations, and manufacturers do not agree. A sample can read within range on one platform and raised on another. An FSH figure is not portable. Comparing this year's number from one lab against last year's from a different one is not a comparison, and any threshold quoted online belongs to whichever analyser produced it.

Luteinising Hormone

LH

A pituitary hormone that triggers ovulation in women and stimulates testosterone production in men.

In men it acts on the Leydig cells, which are the cells that actually make testosterone. In women the mid-cycle LH surge is what releases the egg, and that is what ovulation predictor kits detect.

Read with FSH and testosterone it separates a problem in the testes or ovaries, where LH is high because the pituitary is pushing harder, from a pituitary problem, where LH is low or unhelpfully normal.

In men on testosterone replacement LH is usually suppressed. Why that happens, and why it matters more than it looks, is covered in Sex hormones, in the Body Guide.

Human chorionic gonadotropin is the complication. hCG acts on the LH receptor, so it does LH's job while switching off the body's own LH production. Anyone using it, in fertility treatment or alongside testosterone, will see a suppressed LH that says nothing about pituitary function. Modern LH assays do not mistake hCG for LH, so this is a real biological effect and not a measurement artefact.

Where the numbers come from

No professional body publishes an LH threshold, in either sex. That gap is larger than it sounds. Every major guideline on male hypogonadism instructs clinicians to measure LH and FSH to separate a testicular cause from a pituitary one, and not one of them says at what level. The judgement is made by reading LH against testosterone and against the reference range on that report, and not against a published line.

The population reference range is assay-dependent and varies enormously across the cycle in women. LH is also calibrated against a different international reference preparation from FSH, and manufacturers disagree, so an LH number is not portable between labs.

Biotin. LH is one of the markers biotin affects, and on this test it reads falsely low. Biotin, at the top of this guide, has the detail.

The LH to FSH ratio

A raised LH to FSH ratio was historically associated with polyendocrine metabolic ovarian syndrome (PMOS), previously polycystic ovary syndrome (PCOS). It has not been part of the diagnostic criteria for a long time, the May 2026 rename did not reinstate it, and no guideline endorses a cut-off for it. It is also normal in a large share of people who have the condition, so a normal ratio rules nothing out and a raised one diagnoses nothing. It survives in the research literature as an association and nowhere else.

Prolactin

A pituitary hormone, present and measurable in everyone. Its best known job is milk production, and in pregnancy and breastfeeding it rises enormously. That is the hormone doing exactly what it is for, and not a problem.

Beyond that it has a long list of described roles, in reproductive function, metabolism and immune regulation. Its release is held in check by dopamine, which is why drugs that block dopamine raise it, and there is a surge after orgasm in both sexes thought to be involved in the refractory period afterwards. In men its precise physiological role is less well mapped than in women, and it is not a spare part.

Raised prolactin in men causes low libido, erectile difficulty, low energy and infertility, largely by suppressing the pituitary signals that drive testosterone. It is a standard check in unexplained low testosterone, because a treatable pituitary cause is easy to miss. A markedly raised level can indicate a prolactinoma, a benign pituitary tumour, and very high levels can cause headaches and visual field problems by pressing on nearby structures.

Low prolactin appears to matter too. In middle-aged and older men, low prolactin is associated with worse sexual function, with features of metabolic syndrome and with higher rates of psychological symptoms. In men attending for sexual dysfunction, a low prolactin has also been found to predict later cardiovascular events. The associations are consistent and they are not strong, and nobody has shown that raising a low prolactin changes anything. What they suggest is a range with problems at both ends and not a number to keep as low as possible.

Fatty liver has since been added to that list. Among people with type 2 diabetes, fatty liver is more common the lower the prolactin, and least common in those at the top of the range. In people taking medication to bring a high prolactin down, those pushed below the normal range end up with more weight gain, more insulin resistance and worse cholesterol than those brought back into it. That fits a range with problems at both ends. It does not show that a low prolactin is causing any of them.

In women, outside pregnancy and breastfeeding, raised prolactin causes irregular or absent periods, infertility and sometimes milk production.

It is easily raised by things that have nothing to do with disease, including stress, a rushed journey to the appointment, exercise, sexual activity, nipple stimulation and several common medications. The usual advice is to rest beforehand and avoid exercise and sexual activity for a couple of days. That is sensible, and it is convention and not something a guideline sets out, and it does not remove the problem. The needle itself is enough, because the stress of the blood draw alone can lift prolactin to 2 to 4 times its resting level for up to an hour afterwards.

Where the numbers come from

No professional body publishes a healthy threshold. The population reference range is assay-dependent. Very high results point towards a pituitary cause and are a different matter from the mildly raised results this guide is mostly about.

Because a mildly raised prolactin on a single sample is common and frequently means nothing, the international consensus from the Pituitary Society is to repeat it before acting on it. Where stress is the suspected reason, the repeat is taken through a cannula placed in advance, so that blood is drawn some time after the needle went in.

Two ways the assay misleads

Some people carry a large, biologically inactive complex of prolactin bound to antibody, called macroprolactin, which the immunoassay counts alongside the active hormone. It explains somewhere between 10% and 25% of raised prolactin results, depending on the population and the method, and the person is not unwell.

Macroprolactin is its own test and it is not part of a prolactin result. Hospital labs commonly add it themselves when a prolactin comes back above a set level, usually somewhere around 700 mIU/L, though the level varies between labs and not all of them do it. Private and consumer testing does not work that way. A raised prolactin from a consumer panel has not been screened for macroprolactin, ours included, and if the question matters it has to be ordered separately.

The opposite failure is rarer and more dangerous. A very large prolactin-producing tumour can flood the assay with so much hormone that the chemistry saturates and the result comes back normal or only slightly raised. This is the hook effect, and it is why a large pituitary tumour showing every sign of high prolactin gets its sample re-run diluted. A reassuring prolactin in someone with a known large pituitary mass is a result to question.

Biotin. Prolactin is one of the markers biotin affects, and on this test it reads falsely low. Biotin, at the top of this guide, has the detail.

DHEA-Sulphate

DHEA-S

The sulphated form of dehydroepiandrosterone, an adrenal hormone that acts as a precursor to both testosterone and oestrogen.

It is used as a broad marker of adrenal androgen output. Why the sulphated form is the one measured, why it falls steeply across life, and why it matters far more in women than in men, is all in Adrenal glands, in the Body Guide. This entry is about reading the number.

Where the numbers come from

No professional body publishes a healthy threshold or an optimal level. The population reference range is assay-dependent, and it must be banded by age or it tells you nothing. Check that the one on your own report is.

No professional body publishes a threshold above which further investigation is triggered either. The Society for Endocrinology's 2025 guideline declines to set one, because the data are not robust and DHEA-sulphate methods vary widely between UK and Irish labs. The nearest published figure comes from a review rather than a guideline, and puts a result above 20 µmol/L, about 736 µg/dL in American units, as carrying a high probability of adrenal cancer in postmenopausal women, while noting that in premenopausal women a severe isolated rise can still belong to the renamed ovarian syndrome. A markedly raised result is one to take to a doctor and not to interpret at home. These tumours are rare and the majority of raised results are not one.

Its usefulness in the renamed ovarian syndrome is weaker than usually assumed. Once results are read against age-matched ranges instead of one adult range, only a small proportion of affected women have a genuinely raised DHEA-sulphate. Current international guidance suggests measuring it only where testosterone is not already raised, and notes its poorer specificity.

It is the tiebreaker for an awkward cortisol result. Where a morning cortisol lands in the indeterminate band, neither low enough to indicate adrenal insufficiency nor high enough to exclude it, a DHEA-sulphate read against age helps sort out which way it falls, and a healthy age-adjusted level makes adrenal insufficiency considerably less likely.

That is a rule-out use and it diagnoses nothing.

Supplements. DHEA is sold freely over the counter in the US as a dietary supplement. It cannot be sold that way in the UK, where it is a controlled drug and a prescription-only medicine, and the only licensed British product is a pessary for vaginal dryness. So a reader in this country with a DHEA-sulphate far above the reference range and a bottle in the cupboard has bought from outside the regulated supply chain, with the usual uncertainty about what is actually in it. Whatever the source, a supplement pushes this number up directly, and what the lab then measures is the capsule and not the adrenal gland.

The evidence for taking it is thin in every direction it has been looked at. In healthy adults there is no demonstrated benefit to wellbeing, cognition or sexual function, and the Endocrine Society recommends against routine use. In fertility treatment for women with reduced ovarian reserve, where it is promoted hardest, the randomised trials do not show a live birth benefit. Where any priming treatment has shown one, it was testosterone and not DHEA.

There is no evidence base supporting a target level in a healthy person, which needs saying given how often one is quoted at people.

Biotin. DHEA-sulphate is one of the markers biotin affects, and on this test it reads falsely high. Biotin, at the top of this guide, has the detail.

7-keto-DHEA. Because DHEA itself is prescription-only here, the shelves carry 7-keto-DHEA in its place. It is a different molecule and it is not a controlled drug in the UK, having been considered for control and then deliberately left out of it. The claim made for it is that it is non-hormonal, and on this point the marketing is right, because it is not converted back into DHEA, testosterone or oestrogen and will not raise your sex hormones. Whether it shows up on a DHEA-sulphate result, either by conversion or by confusing the assay, has never been tested by anyone. That is an unknown and not a reassurance, and if you take it, say so when a raised DHEA-sulphate is being interpreted.

Units

The UK reports DHEA-sulphate in µmol/L and the US in µg/dL. Multiply µmol/L by roughly 37. So 10 µmol/L is about 370 µg/dL.

Cortisol

The body's main stress hormone, made by the adrenal glands. It regulates blood sugar, blood pressure, inflammation and the response to physical and psychological stress.

Timing decides what this number means, and there are 2 clocks in play. Cortisol reaches its highest point at or around waking and falls away through the day, so the physiology is anchored to when you got up. Every published reference interval is anchored to the time of day instead. The Roche interval, which sits behind most UK results, is defined for a sample taken between 6am and 10am and says nothing about when the person woke.

Those 2 things come apart in ordinary life. Someone who wakes at 5am and is tested at 9am is 4 hours down the falling limb. Someone who wakes at 8am and is tested at 9am is at the top of it. Both are read against the same range. No guideline addresses this and no correction for it has been published.

What follows is that a sample taken 1 to 3 hours after waking sits in roughly the same part of the curve as the people the reference interval was built from, and for anyone on a conventional schedule it also lands inside that 6am to 10am window. Much earlier than that puts you at or near your own peak, which is higher than where that reference population sat. Get the collection time written on the form either way. An afternoon sample is not a weaker version of a morning one, it is a number with no range to read it against. Why the rhythm exists and why a flattened one is itself a finding is in Adrenal glands, in the Body Guide.

A long fast can raise it, as can acute illness, hard exercise and a stressful journey to the appointment.

Persistently high cortisol is the picture of Cushing's syndrome, and persistently low, of adrenal insufficiency. Where the fault is in the adrenal glands themselves that is Addison's disease. Where it is in the pituitary gland above them it is not, and the 2 behave differently. Both are uncommon and neither is diagnosed on a single morning sample.

Where the numbers come from

NICE published guidance on adrenal insufficiency in August 2024, and it sets the UK bands for a morning sample taken between 8am and 9am. Below 150 nmol/L, adrenal insufficiency is likely. Above 300 nmol/L, it is very unlikely. Between the two is an indeterminate band that needs repeating or a specialist opinion.

The figure of 500 nmol/L appears widely and is frequently misapplied. It comes from the Endocrine Society and it is the peak value after a stimulation test, and not a resting morning sample. The same guideline states in brackets that the cut-off is assay-dependent, and that bracket carries far more weight than its position suggests.

The clearest example of why the assay matters

Cortisol is where the difference between analysers stops being a technicality. The baseline cortisol needed to exclude adrenal insufficiency with full confidence has been put at 336 nmol/L on one manufacturer's assay, 358 nmol/L on another and 506 nmol/L on a third. Same clinical question, same patients, 3 different numbers, and the highest is roughly half as much again as the lowest. That is the whole argument for using the range printed on your own report.

That was tested directly in British labs at the end of 2025. Identical serum specimens went out to several hundred of them, with mass spectrometry target values of 122, 307 and 260 nmol/L. They differed in the number they sent back, and they also differed in the verdict they attached to it, meaning whether the result indicated adrenal insufficiency, excluded it or sat in between. Same blood, different number, different conclusion.

The figures of 150 and 300 also belong to an older generation of assay. The lines that apply on current-generation assays differ from them, and by how much depends on which assay produced the number.

If you take oral oestrogen, this result is not valid

Oestrogen taken by mouth, including the combined oral contraceptive and oral HRT, raises the protein that carries cortisol around the blood. Nearly all the cortisol a standard test measures is the bound fraction, so the total climbs steeply while the free, active hormone does not move at all. The result can look abnormally high when nothing whatever is wrong.

NICE advises stopping oral oestrogen for 6 weeks before a serum cortisol is measured. What it advises in the meantime depends on why the oestrogen is being taken. Where it is for contraception, it says to use another method of contraception. Where it is for HRT, it says to consider switching to a patch or gel. Newer research on the combined pill suggests levels may have settled by 4 weeks. The guidance has not changed, and the same has not been shown for oral HRT. Oestradiol through the skin, as a patch or gel for HRT, does not cause the problem, because it does not pass through the liver first. The combined contraceptive patch is the exception, because the oestrogen in it raises the carrier protein about as much as the pill does.

Biotin. Cortisol is one of the markers biotin affects, and on this test it reads falsely high. Biotin, at the top of this guide, has the detail.

Adrenal fatigue

A large market sells cortisol testing for adrenal fatigue, the idea that chronic stress exhausts the adrenal glands and leaves them unable to produce enough cortisol. It has been looked for properly and it has not been found. The evidence covering exactly the cortisol tests sold for it concluded there is no substantiation that the condition exists. The Endocrine Society says the same and warns that the supplements sold for it may do harm. Fatigue is real and often has a findable cause, and no cortisol test we or anyone else sells will diagnose adrenal fatigue.

What is legitimate beyond a morning blood sample, and what is not

For suspected cortisol excess, two home or outpatient collections are validated as first-line screening tests: late-night salivary cortisol, and a 24-hour urine collection measuring free cortisol. Both need repeating. Overnight dexamethasone suppression is the third first-line option and needs a prescription. None of these is a single blood sample and none is something we offer.

Against that, salivary cortisol profiles sold as a 4-point daily curve, and dried urine steroid panels, are not validated for diagnosing anything. They produce a lot of numbers and a graph, which is some distance from producing an answer.

Anti-Müllerian Hormone

AMH

A hormone produced by the small developing follicles in the ovaries. It gives an estimate of ovarian reserve, meaning roughly how many eggs remain.

What ovarian reserve is and is not, why an antral follicle count is the other half of the assessment, and the unusual thing this marker does in men, are all in Sex hormones, in the Body Guide. This entry is about what the number can be trusted to tell you, and that turns out to be less than it is sold as.

Where the numbers come from

There is no published threshold for this marker. The reference range is strongly age-dependent, and it is assay-dependent to a degree that is unusual even by the standards of this guide.

Your result belongs to the machine that produced it. There is still no international standard for anti-Müllerian hormone. A reference reagent exists, and when it was tested across methods the agreement was poor, with several assays impossible to reconcile with patient samples at all. In practice the major analysers return meaningfully different numbers for the same blood. One widely used platform reads substantially lower than another, and one lab method drifts further from the others the higher the result climbs. A number that looks reassuring on one platform can look borderline on another.

Two things follow. Do not compare an anti-Müllerian hormone from one lab with one from another, and if you are tracking it over time, have it done on the same platform each time or the trend means nothing. Any threshold you find quoted online belongs to whichever assay generated it.

Biotin. Anti-Mullerian hormone is one of the markers biotin affects, and on this test it reads falsely low. Biotin, at the top of this guide, has the detail.

It can be collected on any day of the cycle, which is what makes it easier to get right than FSH or oestradiol. That does not make it perfectly steady. It moves across a single cycle by more than the analytical error alone would explain, so a small change between two tests is not necessarily a change in you.

What lowers it that people do not expect

Hormonal contraception is the big one. Current use of the combined pill lowers anti-Müllerian hormone substantially, and the effect is proportionally largest in women whose level is already at the lower end, which is the worst place for a misleading number. It is also not quick to reverse, with levels still below baseline 3 months after stopping. A woman who comes off the pill and tests a few weeks later is not seeing her own baseline. Hormonal coils do not appear to have the same effect.

Smoking lowers it, in current smokers more than former ones. Ovarian surgery lowers it, and how much depends on the technique used.

Cancer treatment lowers it a great deal, and then it partly comes back. At 12 months after breast cancer treatment it sits around 76% below where it started, and after radioactive iodine for thyroid cancer around 58% below. In every group looked at it rose again somewhere between 6 months and 2 years, and among women who kept ovarian function the long-term trend then followed that of untreated women the same age. An anti-Mullerian hormone taken soon after treatment reads far worse than the eventual position.

What does not independently move it, despite being widely claimed to: body mass index, waist measurement, exercise and alcohol. Vitamin D has been looked at and the effect, if any, is small and uncertain.

What it does and does not predict

Ovarian reserve is not the same thing as fertility, and that gap is what the professional positions are about. Conception depends on egg quality, tubal patency, partner factors and timing, none of which anti-Müllerian hormone measures.

The American College of Obstetricians and Gynecologists says it should generally not be ordered or used to counsel women who are not infertile about their reproductive status or future fertility potential, because using it that way is not supported by high-quality data. NICE guideline NG23 names it in the list of tests not to use to identify perimenopause or menopause in anyone aged 45 or over. The American Society for Reproductive Medicine puts it precisely. Markers of ovarian reserve are good predictors of how many eggs will be retrieved in a treatment cycle, and poor predictors of reproductive potential.

On natural conception the evidence is split. One large independent study of women trying to conceive without treatment found no relationship between a low anti-Müllerian hormone and the chance of getting pregnant. A larger one with commercial ties found a modest relationship at the low end, amounting to a few percentage points of difference in the chance per cycle. Neither supports using it as a personal forecast.

The way it is sold

When the websites selling anti-Müllerian hormone directly to the public were examined, most claimed or implied it indicates the likelihood of conceiving, and only a third stated that it cannot. Women shown accurate information about the test were then considerably less interested in having it than women shown a seller's own material. The demand is partly manufactured by the marketing.

We sell this test. It is useful if you are heading into fertility treatment and want to know what to expect from a stimulation cycle, if you are considering egg freezing, or if premature ovarian insufficiency is being investigated. It is not an egg timer, it will not tell you when your menopause will be, and a low result in a woman with no fertility problem is more likely to frighten her than to inform her.

The ovarian syndrome criterion

Since 2023 a raised anti-Müllerian hormone can stand in for the ultrasound when diagnosing polyendocrine metabolic ovarian syndrome (PMOS), previously polycystic ovary syndrome (PCOS), in adults. One or the other, not both. There is no internationally agreed cut-off for this, and the figures in circulation differ by assay and by population, so the substitution is accepted while the number to use is not.

In men

The Body Guide explains why this marker reads what is happening inside the testis and not in the blood. Two practical consequences follow.

The first is that treatments raising testosterone by different routes do different things to it. Human chorionic gonadotropin, which stimulates the testis itself, drives anti-Müllerian hormone down to normal adult levels. Injected testosterone raises the blood level by a comparable amount and produces only a partial fall, because it does not reproduce the concentration inside the testis.

The second is where it earns a place. In men with unexplained azoospermia being considered for surgical sperm retrieval, a lower anti-Müllerian hormone has been associated with a better chance of finding sperm, where testosterone, inhibin B, age and testicular volume were not. The evidence behind that is inconsistent and the values of men who succeed and fail overlap heavily, so it contributes to the decision without settling it.

An undetectable anti-Müllerian hormone alongside undetectable testosterone indicates absent functional testicular tissue. For general male infertility it adds nothing to routine diagnostics.

And what is not known

Almost all of the work above was done in men with hypogonadotropic hypogonadism being treated to restore fertility, and men taking testosterone for replacement are routinely excluded. So how anti-Müllerian hormone behaves in a man on ordinary testosterone replacement, and whether it is any use for tracking testicular function on treatment, has never been mapped.