How testosterone should actually be tested
Morning, fasting, repeated, and measured on the right assay. Eight ways your number can be wrong, and what to do about each of them.
More men are misdiagnosed by bad measurement than by bad judgment.
That sounds like an overstatement. It isn’t. Testosterone is one of the more difficult things a general laboratory measures. The value changes by the hour, falls after a meal, drops after a poor night’s sleep, and differs measurably depending on which machine ran the sample. A man can produce a result below the diagnostic threshold in the afternoon and a normal result the following morning without anything about his physiology having changed.
The Endocrine Society’s July 2026 statement puts the requirement plainly: a diagnosis should rest on at least two early-morning, fasting testosterone tests, using a standardized assay, in a man who has symptoms consistent with deficiency. Every element of that sentence is there because leaving it out produces wrong answers at a rate that matters.
This page explains how the test works, what makes it unreliable, and how to tell whether your own result was obtained under conditions that make it worth acting on.
Why the time of day matters
Testosterone is secreted in pulses, on a daily rhythm driven by luteinizing hormone from the pituitary. Levels peak in the early morning and fall through the day.
The size of that fall is larger than most men expect. Brambilla and colleagues, writing in the Journal of Clinical Endocrinology & Metabolism in 2009, sampled 66 men aged 30 to 80 across six separate visits — three in the morning, three in the afternoon. In men aged 30 to 40, total testosterone was 20 to 25% lower at 4pm than at 8am. The gap narrowed with age, to about 10% at 70. Among the men who recorded at least one afternoon value below 300 ng/dL (10.4 nmol/L), a substantial share had normal testosterone at every one of their morning visits.
Read that again, because it is the single most important fact on this page. A quarter of the men in that study would have been classified as testosterone deficient or not deficient depending only on when they happened to walk into the phlebotomy room.
The rhythm flattens with age. In the same work, the morning-to-afternoon difference was only around 10% by age 70. It is also blunted in men with primary testicular failure. This is a real nuance — the argument for morning-only testing is strongest in younger and middle-aged men — but it is not a licence to test older men in the afternoon. The reference ranges that define “low” were themselves built from morning samples. Comparing an afternoon value against a morning-derived range is a category error regardless of the patient’s age.
Practically: the draw should be between roughly 7am and 10am. The Endocrine Society, AUA, EAU and the VA/DoD all specify a morning sample; the VA/DoD is most explicit at 6am to 10am, the EAU at 7am to 10am.
Do you need to fast?
Yes, and the reason is more substantial than most patients are told.
A study of 60 men without diabetes, aged 23 to 97, measured what happened to testosterone after a glucose drink and after a mixed meal. After the glucose load, total testosterone fell by an average of 100 ng/dL (3.5 nmol/L) at its lowest point. After the mixed meal, it fell by an average of 123 ng/dL (4.3 nmol/L). The decline began within 20 minutes, bottomed out around 60 minutes, and had still not returned to baseline at two hours.
The consequence: 11% of the men transiently dropped below 300 ng/dL after the glucose load, and 56% did so after the meal. More than half of a group of men, most of whom were not testosterone deficient, would have produced a “low” result if blood had been drawn an hour after breakfast.
The Endocrine Society’s 2018 guideline and its July 2026 statement both require a fasting sample. So do the EAU and the VA/DoD. The AUA guideline specifies early-morning testing but, in the text available to us, does not explicitly require fasting — a small divergence, but one that means two American clinicians following two American guidelines can order the same test under different conditions.
Why one test is never enough
The Endocrine Society’s 2018 clinical practice guideline states it directly: around 30% of men whose initial testosterone falls in the hypogonadal range have a normal testosterone on repeat measurement.
Three men in ten. Not because the first test was performed badly, necessarily, but because testosterone varies within the same person from day to day, and because a single measurement of a fluctuating quantity is a poor estimate of its average.
This is why every major guideline — Endocrine Society, AUA, EAU, VA/DoD — requires two separate measurements before a diagnosis is made. It is also the single most commonly skipped step in commercial testosterone prescribing. A recent analysis presented at ENDO 2026 by researchers at Michigan Medicine, reviewing 200 men who started testosterone therapy, found that only about 12% had a guideline-concordant diagnostic workup — meaning two low morning samples, gonadotropin measurement, and screening for contraindications.
If you were prescribed testosterone on the basis of one blood test, you were prescribed it on the basis of information that is wrong about 30% of the time.
Total, free, and bioavailable testosterone
Most testosterone in the blood is not free to act. It circulates bound to two proteins:
- Sex hormone-binding globulin (SHBG) binds testosterone tightly. Between 40 and 60% of circulating testosterone is bound this way and is generally considered unavailable to tissues.
- Albumin binds testosterone loosely. This fraction dissociates readily and is usually counted as available.
That leaves roughly 1 to 4% genuinely unbound — free testosterone.
Total testosterone is everything: bound plus free. It is what a standard testosterone test measures. Bioavailable testosterone is free plus albumin-bound. Free testosterone is the unbound fraction alone.
For most men, total testosterone is an adequate proxy for what the tissues are seeing, which is why the AUA guideline emphasizes total testosterone and does not routinely recommend free testosterone for initial diagnosis. The proxy breaks down when SHBG is abnormal — and SHBG is abnormal in a large fraction of the men who present asking about testosterone.
SHBG: why total and free testosterone disagree
SHBG is the reason two men with identical total testosterone can have quite different amounts of hormone available to their tissues.
Conditions that lower SHBG — and therefore lower total testosterone while free testosterone may be preserved:
Obesity · insulin resistance and hyperinsulinemia · type 2 diabetes · metabolic-associated fatty liver disease · hypothyroidism · nephrotic syndrome · exogenous androgens · glucocorticoids
The mechanism runs through the liver: hyperinsulinemia and hepatic fat accumulation suppress the transcription factor HNF4α, which reduces SHBG production.
Conditions that raise SHBG — and therefore raise total testosterone while free testosterone may be lower than it looks:
Aging · hyperthyroidism · liver disease including hepatitis C and alcoholic liver disease · estrogens · anticonvulsants · HIV · growth hormone deficiency · low body weight · certain genetic polymorphisms
This produces two clinically important patterns.
The obese man with a low total testosterone. His SHBG is suppressed by insulin resistance. His total testosterone looks deficient. His free testosterone may be entirely normal. Treating him with testosterone treats a number rather than a man — which is why the Endocrine Society’s July 2026 statement says that where obesity with a BMI above 27 is the only identified cause, weight loss is typically first-line therapy rather than testosterone.
The older, thin man with a normal-looking total testosterone. His SHBG is elevated. His total testosterone sits comfortably in range. His free testosterone may be genuinely low, and his symptoms may be genuinely hormonal.
Neither man is well served by looking at total testosterone alone.
When free testosterone actually matters
This is one of the places where the world’s major societies openly disagree, and you should know that before someone quotes a free testosterone number at you as though it settled something.
- The AUA does not routinely recommend free testosterone for initial diagnosis.
- The Endocrine Society uses it for borderline total testosterone results.
- The British Society for Sexual Medicine measures it routinely when total testosterone falls in the 8 to 12 nmol/L (231 to 346 ng/dL) gray zone. The EAU used to say the same; its 2026 edition dropped the gray-zone rule and now calls for free testosterone where SHBG is likely to be altered, which is a narrower instruction.
- Italian endocrine and andrology societies formalize it within their diagnostic algorithm.
- The UK Society for Endocrinology is skeptical of the assays.
- The VA/DoD recommends it when SHBG-altering conditions are present, and is explicit about which method to use.
There is more agreement about how to measure it than about when.
Equilibrium dialysis and equilibrium ultrafiltration are the reference methods. They physically separate the free fraction. They are accurate, slower, and not universally available.
Calculated free testosterone derives the number from total testosterone, SHBG and albumin, most commonly using the Vermeulen equation. It is widely used and reasonably good, but it is a model, not a measurement, and it does not agree perfectly with direct methods. In published comparisons, calculated values have run around 24% higher than measured free testosterone in one cohort, and one multicenter comparison found roughly a 26% difference between younger and older men — meaning the agreement between calculated and measured free testosterone is not uniform across the population you are applying it to.
Direct or analog free testosterone immunoassays should not be used. This is one of the few points on which the guidelines are in unanimous, explicit agreement. The Endocrine Society’s 2018 guideline says clinicians “should not use direct analog-based free testosterone immunoassays, as they are inaccurate.” The EAU says direct free testosterone evaluation with these methods “is not recommended and should be avoided.” The VA/DoD says to avoid the direct analog displacement method.
Despite this, direct analog free testosterone remains on the menu at many laboratories and appears routinely on direct-to-consumer hormone panels. If your free testosterone was measured by a direct immunoassay, the number is not reliable enough to make a treatment decision on.
How the assay changes the answer
Two laboratories can measure the same tube of blood and return different numbers. This is not a scandal; it is a known and quantified property of hormone measurement. But it has consequences at the exact place where it matters most — near the diagnostic threshold.
Immunoassay is fast, automated, cheap, and what most laboratories run by default. It performs least well at the low end of the male range, which is precisely where the diagnosis is made.
Liquid chromatography–tandem mass spectrometry (LC-MS/MS) is more accurate and more specific, particularly at low concentrations. It is the reference-grade method, and the AUA describes it as the gold standard. The VA/DoD’s approach is pragmatic and sensible: run an immunoassay first, and confirm an abnormal result with LC-MS/MS.
How large is the disagreement? A widely cited analysis by the Partnership for Accurate Testing of Hormones, published in Clinical Chemistry, established performance goals for adult male testosterone measurement: imprecision of 5.3%, bias of 6.4%, and total allowable error of 16.7%. Its conclusion was that most available testosterone assays “missed these goals by wide margins.” Even among assays certified as accurate on average, individual samples showed considerable variability.
Sixteen percent total allowable error, as a goal, on a measurement where the difference between 264 and 300 ng/dL changes the diagnosis, tells you how much weight a single number can bear.
CDC certification
The CDC Hormone Standardization Program (HoSt) exists to address exactly this. An assay earns certification by demonstrating mean bias within ±6.4% of the CDC reference measurement procedure across four consecutive quarters, over a measuring range of 2.50 to 1,000 ng/dL. Certification lasts one year and must be renewed.
The Endocrine Society’s July 2026 statement specifically calls for the use of CDC HoSt-certified assays, noting that many testosterone tests otherwise give widely variable results. The CDC publishes the list of certified assays. It is reasonable to ask your laboratory whether the assay used on your sample is on it. Most patients never ask, and most clinicians never check.
Why laboratories report different reference ranges
You may have seen one laboratory report a normal range of 300 to 1,000 ng/dL and another 240 to 870. Neither is lying. Reference ranges are typically derived from whatever population that laboratory or manufacturer sampled, which varies in age, body weight and health.
The attempt to fix this is worth knowing about. A 2017 study in the Journal of Clinical Endocrinology & Metabolism by Travison and colleagues harmonized testosterone measurements across cohorts — 456 men from the Framingham Heart Study third generation and 729 from a Belgian cohort, 1,185 healthy non-obese men in total — using morning fasting samples measured by LC-MS/MS cross-calibrated to the CDC reference method.
For men aged 19 to 39, the harmonized reference range was 264 to 916 ng/dL (9.2 to 31.8 nmol/L).
Two things in that data deserve attention.
First: the 2.5th percentile was 264 ng/dL. That is the origin of the 264 ng/dL threshold used by the VA/DoD. It is not an arbitrary number; it is the bottom 2.5% of healthy young men.
Second, and more useful to you: the median was 531 ng/dL (18.4 nmol/L). Half of healthy young men were below that. A great deal of marketing is built on the implication that a testosterone in the 400s is a problem requiring correction. In a healthy young reference population, a substantial share of men live there.
Why your testosterone result may be wrong
Eight ways a number can mislead you, with the approximate size of each error where it has been measured.
1. It was drawn in the afternoon. Up to 20–25% lower than the same man’s morning value if he is under 40; around 10% lower at 70.
2. You weren’t fasting. Average falls of 100 ng/dL after a glucose load and 123 ng/dL after a mixed meal, with more than half of men in one study transiently dropping below 300 ng/dL after eating.
3. It was only done once. Roughly 30% of initially low results are normal on repeat.
4. You were sleep-deprived. In a study of ten healthy young men, one week of five hours’ sleep per night lowered daytime testosterone by 10 to 15% — a change the authors compared to ten to fifteen years of normal aging.
5. You were acutely ill. Systemic illness suppresses the hypothalamic-pituitary-gonadal axis, and the degree of suppression tracks with severity. A testosterone measured during or shortly after significant illness is not a measure of your baseline.
6. The assay was an immunoassay at the low end of the range. Where accuracy is worst and the diagnosis is made.
7. Your free testosterone was measured by direct immunoassay. Every major guideline recommends against this method.
8. You were taking biotin. This one is almost never mentioned and it produces errors in the dangerous direction. High-dose biotin — common in hair, skin and nail supplements, and used therapeutically at much higher doses — interferes with competitive immunoassays in a way that produces falsely elevated testosterone results. Guidance from the American Association for Clinical Chemistry suggests an eight-hour washout for typical supplement doses of 5 to 10 mg, and up to 72 hours for high-dose biotin therapy, longer if kidney function is impaired. A man taking a hair supplement can be told his testosterone is fine when it is not.
One that works the other way
Recent hard exercise raises testosterone, it does not lower it. A meta-analysis in the Journal of Endocrinological Investigation found that acute moderate-to-high-intensity exercise significantly increases testosterone immediately afterward, with the effect largely gone by 30 minutes. Mild activity did nothing.
So the common advice to “avoid exercising before your test” is directionally right but for the wrong reason. Training hard shortly before a draw risks a falsely reassuring result, not a falsely low one. If you want to know your true baseline, don’t lift on the way to the lab.
Home tests and saliva tests
Here we have to be straight with you about what we don’t know.
At-home finger-prick and dried blood spot tests. Dried blood spot collection paired with LC-MS/MS analysis has been validated as a method — a 2024 analytical paper published such a validation. That establishes the technique can work. It does not establish that any particular consumer product does work.
We looked for a systematic evaluation of the direct-to-consumer testosterone testing kits currently on the market, measured against venous LC-MS/MS. We could not find one. That is a genuine gap in the literature, not a gap in our search, and we would rather tell you that than quote an accuracy figure we cannot support.
What we can say with confidence is that the collection conditions are usually the bigger problem than the assay. A home kit does not enforce a 7am draw, does not enforce fasting, and does not enforce a repeat. Even a perfectly accurate assay applied to a sample collected at 3pm after lunch produces a number you should not act on.
Saliva testing. Salivary testosterone tracks the free rather than the total fraction, and a 2014 critical evaluation raised substantive questions about its clinical reliability. No major guideline recommends salivary testosterone for diagnosing testosterone deficiency. It has a legitimate role in some research settings. It does not have one here.
A home test is reasonable as a prompt to seek proper evaluation. It is not a basis for a diagnosis, and it should never be the basis for a prescription.
What level is actually too low?
There is no single answer, and anyone who gives you one without qualification is simplifying past the point of usefulness.
| Source | Threshold for total testosterone |
|---|---|
| VA/DoD (January 2026) | 264 ng/dL (9.2 nmol/L) |
| Endocrine Society (July 2026 statement) | “near 300 ng/dL” (10.4 nmol/L) |
| AUA | 300 ng/dL (10.4 nmol/L) |
| EAU (2026) | 12 nmol/L (346 ng/dL) or below is the diagnostic threshold; treatment benefit concentrates below 8 nmol/L (231 ng/dL); below 6 nmol/L (173 ng/dL) prompts pituitary imaging |
A published cross-society analysis in 2025 found thresholds spanning 8 to 12 nmol/L — roughly 231 to 346 ng/dL — across guidelines.
The practical consequence is worth stating plainly: a man with a testosterone of 320 ng/dL is normal under the AUA and the Endocrine Society, and potentially hypogonadal under the EAU. He has not changed. The threshold has.
This is not a failure of medicine. Testosterone concentrations in men are a continuous distribution with no natural break separating “deficient” from “sufficient,” and any threshold is a decision about where to accept the tradeoff between treating men who don’t need it and missing men who do. Different societies have weighed that tradeoff differently, with different populations in mind.
What follows from it is a discipline rather than a number. A threshold is a prompt to think, not a trigger to prescribe. A man at 250 ng/dL with no symptoms does not obviously need treatment. A man at 340 ng/dL with unambiguous symptoms, high SHBG and a low free testosterone might genuinely have a hormonal problem. The number opens the conversation; it does not end it.
We cover the disagreement itself, and what drives it, in Why Experts Don’t Always Agree.
What good testing looks like
If you are being evaluated properly, this is roughly what it should look like:
- Symptoms first. Testing without a clinical reason produces incidental findings and anxiety. The Endocrine Society explicitly says there is insufficient evidence to support population-level screening of asymptomatic men.
- Total testosterone, drawn between 7 and 10am, fasting.
- A second confirmatory sample, on a different day, under the same conditions.
- SHBG, and free testosterone where SHBG is likely to be abnormal or the total sits in the gray zone — calculated from a reliable equation, or measured by equilibrium dialysis. Not by direct immunoassay.
- LH and FSH, to distinguish a testicular problem from a central one. This step separates a workup from a number.
- Whatever else the history demands — prolactin, thyroid function, iron studies, a medication review, an assessment for sleep apnea.
- A conversation about cause before any conversation about treatment.
If steps 3 and 5 were skipped, you have a testosterone level. You do not yet have a diagnosis.
What remains uncertain
Honest limits on everything above.
How much within-person variation is normal. We know testosterone varies day to day. The published estimate of intra-individual biological variation is not something we can state with confidence, and it matters — it determines how far apart two measurements can be before the difference means something.
Whether morning testing matters equally at every age. The diurnal rhythm is clearly blunted in older men. Whether that justifies relaxing the morning requirement after 70 has not been settled, and no guideline currently does so.
How much immunoassays actually disagree with LC-MS/MS at the specific concentrations where diagnosis happens. Published work confirms the problem is real and worst at the low end. We were not able to verify a specific bias figure at the 200–350 ng/dL range, and we are not going to invent one.
Whether direct-to-consumer test kits are accurate. No systematic evaluation of current products against venous LC-MS/MS was found. This is an evidence gap, and given how many men now enter the healthcare system through one of these kits, it is an important one.
Where the threshold should actually be. The disagreement between societies is not a labelling dispute. It reflects genuine uncertainty about which men benefit from treatment, and it will not be resolved by better assays.
How long after an illness a testosterone measurement becomes valid again. Suppression during acute illness is well documented. The recovery time course is not well characterized.
Questions patients ask
You can test testosterone at any time of day.
No. Testosterone follows a daily rhythm, and afternoon values can be substantially lower than morning values in the same man.
What the evidence showsBrambilla and colleagues (*JCEM* 2009) sampled 66 men across six visits. Men aged 30 to 40 had total testosterone 20 to 25% lower at 4pm than at 8am; by 70 the gap was about 10%. Several men with a sub-300 ng/dL afternoon value were normal at every morning visit.
What remains uncertainThe rhythm flattens with age — around 10% by 70 — and whether older men can be tested later in the day without penalty has not been formally settled.
Bottom lineGet tested between 7 and 10am, at any age.
Strong
You don't need to fast for a testosterone test.
You do. Eating lowers testosterone measurably, within an hour.
What the evidence showsIn 60 men without diabetes, testosterone fell an average of 100 ng/dL after a glucose load and 123 ng/dL after a mixed meal. More than half the men transiently dropped below 300 ng/dL after eating.
What remains uncertainWhether the drop is proportionally similar in men who are already deficient is less well studied.
Bottom lineFast overnight. It is the easiest way to avoid a false diagnosis.
Strong
One low testosterone result means you have low testosterone.
It doesn't. About 30% of initially low results are normal when repeated.
What the evidence showsThe Endocrine Society's clinical practice guideline states that around 30% of men with an initial testosterone in the hypogonadal range have a normal level on repeat measurement. Every major guideline requires two samples.
What remains uncertainHow many repeats are optimal in men whose two results straddle the threshold.
Bottom lineIf you were diagnosed on one test, you were not adequately evaluated.
Strong
If my first test was low and my second was normal, the second must be wrong.
More often it's the first that was misleading, though neither is definitive on its own.
What the evidence showsTestosterone varies within an individual from day to day, and regression to the mean means an initially extreme value tends to be less extreme on retest. This is the expected pattern, not an anomaly.
What remains uncertainWith two discordant results, the right next step depends on the symptoms, the gap between the values, and whether conditions were controlled — there is no formula.
Bottom lineDiscordant results mean you need a clinician's judgment, not a third opinion from the internet.
Moderate
One bad night's sleep will tank your testosterone.
One night matters less than a week, but sustained sleep restriction has a substantial effect.
What the evidence showsIn ten healthy young men, one week of five hours' sleep per night lowered daytime testosterone by 10 to 15% — a change the authors likened to ten to fifteen years of aging.
What remains uncertainThe study was small and in young men. The effect of a single disrupted night is less well quantified.
Bottom lineDon't test after a run of bad nights. And chronic sleep restriction is a plausible cause of a genuinely low result.
Moderate
Exercising before your test will make your testosterone look low.
The opposite. Hard exercise raises testosterone acutely.
What the evidence showsA meta-analysis found acute moderate-to-high-intensity exercise significantly increases testosterone immediately afterward, with the effect largely dissipated by 30 minutes. Mild activity had no effect.
What remains uncertainWhether repeated heavy training over weeks lowers baseline testosterone is a different question with a different answer.
Bottom lineTraining before a blood draw risks a falsely reassuring number, not a falsely low one.
Moderate
Saliva testosterone testing is accurate.
Not accurate enough to diagnose deficiency.
What the evidence showsSalivary testosterone reflects the free rather than the total fraction, and a 2014 critical evaluation raised significant questions about its clinical reliability. No major guideline recommends it for diagnosis.
What remains uncertainIt has legitimate research applications, and assay methods continue to develop.
Bottom lineNot a basis for diagnosis or treatment.
Limited
Home testosterone tests are accurate.
We don't actually know, and the collection conditions are usually the bigger problem.
What the evidence showsDried blood spot collection with LC-MS/MS analysis has been validated as a technique. We could not locate any systematic evaluation of current consumer kits against venous LC-MS/MS.
What remains uncertainEssentially the whole question. Given how many men now enter care through a home kit, this is a meaningful gap.
Bottom lineUseful as a prompt to get properly evaluated. Not a diagnosis.
Limited
Free testosterone is more important than total testosterone.
More informative in specific situations, not universally better.
What the evidence showsSocieties disagree. The AUA does not routinely recommend it for initial diagnosis; the BSSM uses it routinely in the 8–12 nmol/L gray zone; the EAU dropped that rule in 2026 in favour of measuring where SHBG is likely altered; the Endocrine Society uses it for borderline results. Free testosterone becomes genuinely important when SHBG is abnormal.
What remains uncertainThe threshold for low free testosterone is less well established than for total, and calculated values don't agree perfectly with measured ones.
Bottom lineUseful when SHBG is abnormal or total testosterone is borderline. Not a routine replacement for total.
Moderate
SHBG doesn't really matter.
SHBG is often the entire explanation for a confusing testosterone result.
What the evidence showsSHBG binds 40 to 60% of circulating testosterone. Obesity, insulin resistance, type 2 diabetes and hypothyroidism lower it; aging, hyperthyroidism, liver disease and estrogens raise it. Either direction can make total testosterone misrepresent what tissues are exposed to.
What remains uncertainHow often total and free testosterone genuinely lead to different clinical decisions has not been well quantified.
Bottom lineIf your total testosterone doesn't match how you feel, SHBG is the first thing to check.
Strong
You can have low total testosterone because SHBG is high.
High SHBG raises total testosterone. It's low SHBG that depresses it.
What the evidence showsSHBG carries most circulating testosterone, so more binding protein means a higher total. High SHBG typically produces a normal or high total testosterone alongside a low free fraction — the trap being a total that looks reassuring.
What remains uncertainHow symptomatic men with high SHBG and low free testosterone respond to treatment is less studied than the classic presentation.
Bottom lineHigh SHBG hides deficiency behind a normal-looking total. Low SHBG creates deficiency that isn't there.
Strong
Total testosterone can't be normal if free testosterone is low.
It can, and this is exactly the pattern that gets missed.
What the evidence showsIn men with elevated SHBG — older, thin, hyperthyroid, on anticonvulsants, or with liver disease — total testosterone can sit in the reference range while the free fraction is genuinely low.
What remains uncertainWhether these men benefit from treatment as reliably as men with low total testosterone is not well established.
Bottom lineA normal total testosterone does not rule out deficiency when SHBG is high.
Moderate
300 ng/dL is the cutoff for low testosterone.
It's one cutoff, used by two American bodies. Others use different ones.
What the evidence showsThe VA/DoD uses 264 ng/dL. The Endocrine Society's July 2026 statement cites a common clinical threshold near 300. The AUA uses 300. The EAU treats below 12 nmol/L (346 ng/dL) as likely hypogonadal with a gray zone down to 8 nmol/L. A 2025 cross-society analysis documented a spread of 231 to 346 ng/dL.
What remains uncertainWhich threshold best identifies men who benefit from treatment. This is not an assay problem and better measurement will not settle it.
Bottom lineThe same man can be deficient in Europe and normal in a US veterans' clinic. Treat the threshold as a prompt to think.
Strong that guidelines differ; Limited on which is right
If my testosterone is 400 I should still consider treatment.
400 ng/dL is unremarkable in a healthy population, and no guideline supports treating it.
What the evidence showsIn the harmonized reference range derived from 1,185 healthy non-obese men, the median for men aged 19 to 39 was 531 ng/dL, with the 5th percentile at 303. A value of 400 sits comfortably within the normal distribution.
What remains uncertainWhether a small subset of men with normal levels have clinically meaningful androgen resistance is an open question with no validated way to identify them.
Bottom lineIf your testosterone is 400 and you feel unwell, the cause is probably not your testosterone.
Strong
Different labs give different results because one of them is wrong.
Both can be operating within accepted performance limits and still disagree.
What the evidence showsAnalysis by the Partnership for Accurate Testing of Hormones set desirable performance goals of 5.3% imprecision, 6.4% bias and 16.7% total allowable error — and concluded most available assays missed those goals by wide margins. Reference ranges also differ because they're derived from different populations.
What remains uncertainThe precise magnitude of disagreement at the low concentrations where diagnosis happens.
Bottom lineUse the same lab for serial measurements, and ask whether the assay is CDC-certified.
Strong
LC-MS/MS is better than immunoassay.
Yes, particularly at low concentrations — which is where it matters.
What the evidence showsThe AUA describes LC-MS/MS as the gold standard. Immunoassays perform least well at the low end of the male range. The VA/DoD recommends confirming an abnormal immunoassay result with LC-MS/MS.
What remains uncertainWhether routine first-line LC-MS/MS for everyone is worth the cost, versus confirming abnormal results.
Bottom lineA borderline or low immunoassay result deserves confirmation by mass spectrometry.
Strong
Testosterone can swing wildly from one day to the next.
It varies meaningfully day to day, though "wildly" overstates it.
What the evidence showsAround 30% of initially low results normalize on repeat, which reflects real day-to-day variation plus regression to the mean. Time of day and day-to-day variation both contribute; the studies that measured them did not sample in a way that cleanly separates the two.
What remains uncertainThe precise intra-individual biological variation is not something we can state confidently.
Bottom lineEnough variation to require two tests. Not so much that testing is futile.
Moderate
Every man over 40 should have his testosterone checked.
No. Testing without symptoms creates problems rather than solving them.
What the evidence showsThe Endocrine Society's July 2026 statement says there is insufficient evidence to support population-level screening of asymptomatic men. Screening a population with a continuous distribution against an arbitrary threshold generates findings that prompt treatment without demonstrated benefit.
What remains uncertainWhether specific high-risk groups — men with type 2 diabetes, HIV, or on long-term opioids or glucocorticoids — warrant case-finding is a more open question than general screening.
Bottom lineTest because you have symptoms or a specific risk factor, not because of your birthday.
Moderate
Supplements can't affect your testosterone test.
Biotin can, and it pushes the result in the reassuring direction.
What the evidence showsHigh-dose biotin interferes with competitive immunoassays and produces falsely elevated testosterone. Guidance from the American Association for Clinical Chemistry suggests an eight-hour washout for 5 to 10 mg doses and up to 72 hours for high-dose therapy, longer with impaired kidney function.
What remains uncertainHow often this actually changes clinical decisions in practice hasn't been quantified — but biotin is in a great many hair, skin and nail products.
Bottom lineTell the lab what supplements you take, and stop biotin before testing.
Strong
A testosterone test alone can diagnose low testosterone.
No. A testosterone level is one input to a diagnosis, not the diagnosis.
What the evidence showsEvery major guideline requires symptoms plus consistently low, correctly measured testosterone. The Endocrine Society's July 2026 statement is explicit that symptoms alone are not diagnostic — and the converse holds equally. LH and FSH are needed to locate the problem, and reversible causes need excluding before treatment.
What remains uncertainNothing about the principle. Considerable uncertainty about the thresholds within it.
Bottom lineA number is information. A diagnosis is an explanation.
Strong
Where to go next
If your result was low and correctly obtained, the next question is not what treatment but why: Why Is Testosterone Low?
If you want to understand a specific value on your report: The Lab Library
If you have been offered testosterone: Should You Start Testosterone?
If your number is low and something else might explain how you feel: When Testosterone Isn’t the Answer
Ready for testosterone care built on a diagnosis?
The first step is a full endocrine evaluation, not a prescription. We see patients across San Diego County and welcome referrals from other physicians.