Total testosterone: the number the whole evaluation rests on
One value is not a diagnosis. How much the clock, the meal and the assay move it, where 264 to 916 ng/dL actually comes from, and why no two guidelines agree on when to draw it.
This is the test the whole evaluation rests on, and it is less stable than the report makes it look.
What it measures
Total testosterone is every testosterone molecule in the sample, whatever it is attached to. Most of it is bound — tightly to sex hormone-binding globulin, loosely to albumin — and a small fraction circulates unbound.
You will see that split given as precise percentages — around 44% to SHBG, a little over half to albumin, a couple of percent free. Those figures circulate widely and we have not been able to trace them to a readable primary measurement. They descend from a 1981 paper that computed the distribution from a mathematical equilibrium model rather than measuring it, whose numeric table we have not read; the versions in current circulation come from reviews and a professional-society white paper, and they do not entirely agree with one another. A 2017 review by several of the field’s leading investigators argued that the underlying model’s assumptions about stoichiometry and binding affinity “are not supported by published experimental data and are most likely inaccurate.” What has been measured: when free testosterone is determined by equilibrium dialysis rather than calculated, the median was 1.5% of total, not 2%.
The more important point is that the split is not a constant. It is set by the man’s own SHBG, which differs several-fold between men — cohort means run around 42 to 44 nmol/L with standard deviations of 19 to 22. That is the entire subject of the SHBG page, and it is why two men with an identical total testosterone can have very different amounts available to tissue.
Why it matters
Every guideline in current use starts here. The Endocrine Society, the AUA, the EAU and the VA all make total testosterone the initial diagnostic test. Three of them — the Endocrine Society, the EAU and the VA — position free testosterone as a second-line test used only when binding is in question; the AUA makes no graded statement on free testosterone at all. Whatever its weaknesses, this is the measurement the thresholds are attached to.
When it’s ordered
For symptoms or signs that could reflect testosterone deficiency — reduced libido, erectile difficulty, loss of morning erections, gynecomastia, small testes, infertility, loss of body hair. The AUA also recommends measuring it, even without symptoms, in men with unexplained anemia, bone density loss, diabetes, prior chemotherapy or testicular radiation, HIV, chronic opioid use, male infertility, pituitary disease, or chronic corticosteroid use.
The single most important thing about this test
One value is not a diagnosis, and the reasons are quantifiable.
Time of day. In a study that sampled 66 men across six separate visits — three mornings, three afternoons — total testosterone in men aged 30 to 40 was 20 to 25% lower at 4pm than at 8am. The gap narrowed with age to about 10% at 70. Of 17 men who had at least one afternoon value below 300 ng/dL, nine were normal at all three morning visits, including four of the nine men aged 66 to 80. Note what that last figure rules out: the common claim that older men no longer need a morning draw. The authors of that study concluded the opposite, explicitly recommending morning measurement “in both young and older men.”
Food. A standard 75-gram glucose load dropped mean testosterone by 25%, and 10 of 66 men with a normal fasting value fell into the hypogonadal range at some point during the test. An ordinary 550-calorie mixed meal produced a 30% decline at 60 to 120 minutes. In both studies SHBG and LH were unchanged, so this is a real fall in the hormone, not a shift in binding.
Sleep. One week of five-hour nights in ten healthy young men lowered daytime testosterone by 10 to 15%. The authors described it, in their own words, as a “small convenience sample” — but the comparison they drew is worth keeping: normal aging costs 1 to 2% per year.
Alcohol. In eight young men given 1.5 g/kg of ethanol over an evening, testosterone fell by 23% — and the timing matters more than the number. The fall occurred 10 to 16 hours after drinking began, which lands squarely in the next morning’s draw window. “Don’t drink the night before your blood test” is better advice than it usually sounds.
Illness. Within 24 hours of a head injury, a myocardial infarction, or elective surgery, mean testosterone fell by 271, 202 and 195 ng/dL respectively — decreases of 55%, 43% and 58% — with further decline to mean nadirs of 93 and 117 ng/dL after brain injury and after elective surgery respectively, and full recovery as the men recovered. In a separate series, 62% of acutely admitted men had testosterone suppressed below the normal range. A panel drawn during an illness describes the illness.
Drugs. Men on sustained-release oral opioids had subnormal total testosterone in 74% of cases, in a dose-related pattern; every one of the 28 men taking 100 mg or more of methadone equivalent daily had a subnormal testosterone or estradiol. Chronic glucocorticoids reduced testosterone in 14 of 16 men to a mean of 211 ng/dL, against 449 in 11 matched controls, inversely related to steroid dose and with binding proteins unaffected.
Random biological variation, which is bigger than all of the above. Repeated sampling of the same men gives a within-person coefficient of variation of about 10 to 11% in two independent studies, but a study designed around serial morning draws in 87 healthy men found a between-day within-person CV of 18.7% and concluded that a change of 52% between two morning results is needed before you can call it real. The same study made the point more bluntly than we would have: a man’s own set point “could decrease by half and still be within the reference interval.”
That study is also the corrective to the emphasis of this whole section. It measured the afternoon fall at 43 ng/dL and the ordinary spread between two morning samples at about 115 ng/dL — and concluded that the diurnal effect “was dwarfed by the normal biological variation observed for repeat morning samples.” Timing, fasting and assay are all worth controlling, and they are the parts you can control. But the largest single reason to repeat a low testosterone is not that the first one was drawn badly. It is that testosterone is a noisy quantity and one measurement is a poor estimate of it.
Evidence: Established.
Where 264 to 916 ng/dL comes from
This is the most-quoted range in the field and it is more specific than its use suggests.
It comes from a 2017 harmonization study that took 9,054 men from four cohort studies in the United States and Europe, cross-calibrated their assays against the CDC reference method, and generated a common scale. In healthy, non-obese men aged 19 to 39, the harmonized 2.5th to 97.5th percentiles were 264 to 916 ng/dL (9.2 to 31.8 nmol/L).
Three conditions travel with that number and are usually dropped. It describes men 19 to 39, not adult men generally. It describes men with BMI under 30 and no cancer, cardiovascular disease, diabetes, hypertension or hyperlipidemia. And it is only meaningful on an assay traceable to the CDC reference method — applied to an uncertified immunoassay it is a number from one scale being read against another. The same paper concluded that a substantial share of the differences between cohorts was caused by assay differences alone.
The assay problem
Two laboratories can measure the same serum and disagree by more than the distance between published diagnostic thresholds.
In the study that produced the most-quoted figures, six immunoassays were compared against mass spectrometry in 122 men. One platform ran a mean of 90 ng/dL low; another ran 99 ng/dL high. Just over 60% of samples in the adult male range landed within ±20% of the mass-spectrometry result. That study is from 2004 and single-centre, and it should be cited as such — it predates the CDC standardization program entirely. A 2018 replacement study across five modern platforms found an average bias in men of 11.6%, recovery of a reference standard ranging from 79% to 149%, and one widely used platform with a functional sensitivity of about 101 ng/dL — meaning it cannot reliably distinguish among the values that actually decide a diagnosis.
Mass spectrometry is the reference method, and the EAU and AUA both say so. The AUA adds the realistic qualifier that not every laboratory has it. The VA takes the most practical position of the four: an immunoassay is acceptable for the initial measurement, but an abnormal result “should then always be confirmed by liquid chromatography tandem mass spectrometry.”
On CDC certification. The certification standard is a mean bias within ±6.4% of the CDC reference method, sustained across four consecutive quarters, and it applies to a specific method, reagent lot, calibrator lot and instrument. It is not a guarantee about your specimen. The analysis that set the 6.4% figure noted that among certified assays, “individual sample measurements could show large variability in terms of precision, bias, and total error.” Certification is a claim about a laboratory’s average, not about your result.
Nobody agrees on when to draw it
This is worth seeing laid out, because it is usually presented as settled.
| Source | Threshold | Timing |
|---|---|---|
| Endocrine Society (2018) | Harmonized range 264–916 ng/dL cited; no single cutoff issued | “Morning,” fasting, on two separate mornings — no clock hours given |
| AUA (2018, validity confirmed 2024) | Below 300 ng/dL, Moderate Recommendation, Evidence Grade B | “Early morning,” two occasions (Strong, Grade A) — no clock hours, and no explicit fasting requirement |
| EAU (2026) | 12 nmol/L (346 ng/dL), Strong | 07:00 to 10:00, fasting, Strong. The 2024 edition said 07:00 to 11:00 |
| VA, Recommendations for Use (January 2026) | 264 ng/dL, framed as derived from the Endocrine Society and conditional on CDC certification; below 150 ng/dL triggers evaluation for secondary hypogonadism | 6am to 10am, fasting, confirmed on two occasions |
No two of them specify the same window, and the two American bodies specify none at all. If you have been told “the Endocrine Society says draw before 10am,” that is an interpolation — a sensible one, but not a quotation. Note also that the AUA’s 300 ng/dL figure carries Grade B evidence, while the requirement for two morning samples carries Grade A. The procedure is better supported than the number.
Our own practical instruction, and the reason we give it: draw between 7 and 10am, fasting, twice. That window satisfies every guideline simultaneously.
What a high value can mean
Exogenous testosterone, first and most often. Then: a high-SHBG state — aging, hyperthyroidism, cirrhosis, enzyme-inducing anticonvulsants — where total testosterone can be normal or high while free testosterone is not; hCG or a selective estrogen receptor modulator; rarely an androgen-secreting tumour; and an assay running high, which the numbers above should make you take seriously as a differential rather than a technicality.
What a low value can mean
That question has its own chapter, because the answer determines everything that follows: see why is testosterone low. The short version is that a low total testosterone is a finding, not a diagnosis, and a meaningful share of the causes are reversible.
How it relates to treatment
Before starting: two morning fasting values, on an assay you can name, with LH and FSH to locate the problem and SHBG where binding is in doubt.
During treatment: the VA sets a ceiling, recommending dose reduction if total testosterone exceeds 900 ng/dL. Timing of the draw relative to the dose matters more than most schedules acknowledge and varies by formulation — that belongs with the treatment chapter.
What it does not do: total testosterone on treatment does not tell you whether the man feels better. Symptoms and the number are only loosely coupled, and titrating to a target number rather than to a person is the commonest failure mode in this field.
Related
The Lab Library · How testosterone should actually be tested · SHBG · Free testosterone
Questions patients ask
My testosterone was 280, so I have low testosterone.
You have one number. Whether it is a diagnosis depends on when it was drawn, what you had eaten, which assay ran it, and what the second one says.
What the evidence showsA 75-gram glucose load dropped mean testosterone 25% and pushed 15% of men with normal fasting values into the hypogonadal range. An afternoon draw runs 20 to 25% below a morning draw in men aged 30 to 40. One immunoassay platform runs about 90 ng/dL low against mass spectrometry. And in serial morning sampling, a 52% change is needed before a difference counts as real.
What remains uncertainNothing about the direction of these effects. What is uncertain is how much each contributed to your particular value — which is why the answer is a second test, not an argument.
Bottom lineRepeat it properly before anyone acts on it.
Strong
I'm over 60, so the morning rule doesn't apply to me.
The rule applies at every age. The effect gets smaller; it does not go away.
What the evidence showsDiurnal variation fell from 20 to 25% in men aged 30 to 40 to about 10% at 70. But among men aged 66 to 80 who had a low afternoon value, four of nine were normal at all three morning visits — and the study's own conclusion was to restrict testosterone measurement to morning hours "in both young and older men."
What remains uncertainNothing material.
Bottom lineMorning, fasting, at any age.
Strong
264 to 916 ng/dL is the normal range for men.
It is the normal range for healthy non-obese men aged 19 to 39, on a CDC-traceable assay.
What the evidence showsThe range comes from a harmonization of four cohorts totalling 9,054 men, and the published figure is explicitly for the 19-to-39, BMI-under-30, disease-free subgroup. The same paper found that a substantial part of the variation between cohorts was caused by assay differences.
What remains uncertainWhat the right range is for a 65-year-old — which is the question the field has been arguing about for two decades and has not settled.
Bottom lineUseful, real, and narrower in scope than its use.
Moderate
A CDC-certified assay means my result is accurate.
It means the method's average is accurate. Your individual specimen is a separate question.
What the evidence showsCertification requires a mean bias within ±6.4% of the reference method across four consecutive quarters, tied to a specific instrument, reagent lot and calibrator lot. The analysis that derived that standard observed that even among certified assays, individual sample measurements could show large variability in precision, bias and total error.
What remains uncertainHow large that per-specimen spread is on any particular platform today.
Bottom lineCertification is worth having and is not a per-result guarantee.
Moderate
My testosterone is fine — it's in the reference range.
More complicated than it sounds, and this argument is used in both directions.
What the evidence showsBecause within-person variation is smaller than between-person variation, a man's own set point could fall by half and still sit inside the population interval — that is a direct quotation of the finding, from a study of 87 healthy men. The same fact is the standard justification for treating men who are "in range but declining."
What remains uncertainWhether within-individual decline predicts benefit from treatment. No trial has tested it, which is why we grade the treatment inference low even though the measurement fact is solid.
Bottom lineThe observation is real. The therapeutic conclusion drawn from it is not established.
Limited
The guidelines all say to test between 7 and 10 in the morning.
They do not. Two of the four specify no hours at all.
What the evidence showsThe Endocrine Society and the AUA say "morning" and "early morning" respectively, with no clock times. The VA says 6am to 10am. The EAU's 2026 edition says 07:00 to 10:00 — its 2024 edition said 07:00 to 11:00.
What remains uncertainNothing; this is a documentary matter.
Bottom line7 to 10am fasting satisfies all of them, which is why we recommend it — but it is our synthesis, not a quotation.
Strong
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.
