Should You Get Your Testosterone Tested? The Guideline Says No — In One Sentence
On 15 July 2026, the US Department of Defense announced that every service member aged 30 and over will have their testosterone measured once a year. It gets added to the health assessment they already do. Younger troops can opt in. Treatment, if a doctor recommends it, stays voluntary. [8][9]
It is worth being precise about what that is and is not. Nobody is being put on testosterone. This is a screening programme — testing people who have not complained of anything, to see whether something turns up.
Which is the exact thing the governing clinical guideline recommends against, in one sentence, at its strongest grade.
What the guideline actually says
The Endocrine Society’s clinical practice guideline on testosterone is the reference document American endocrinologists work from. Its recommendation 1.2 reads, in full:
“We recommend against routine screening of men in the general population for hypogonadism.” [1]
That is a Grade 1 recommendation — the category the Society uses for “we recommend” rather than “we suggest”. It is not a hedge or a footnote.
The recommendation immediately before it explains the logic. A diagnosis requires symptoms and signs of testosterone deficiency and unequivocally, consistently low readings. [1] Both. Not a number on its own.
The guideline is blunt about why. Its technical remark: “Low T concentrations occur frequently without symptoms or signs of testosterone deficiency, and these low levels (alone) do not establish a diagnosis of hypogonadism.” [1]
And on screening specifically: “screening for hypogonadism does not fulfill the necessary criteria to justify population-level screening,” followed by the line that ought to end the argument — “No clinical trials have assessed the effectiveness of screening strategies.” [1]
Nobody has tested whether testing helps. That is not an oversight anyone is hiding; it is stated plainly in the document.
So why is a blood test controversial?
Because this particular number is far less solid than a number printed on a lab report looks.
| Drinking a glucose drink | −25% |
| … and the share of normal men it pushed into the “low” range | 15% |
| Time of day — why the guideline says morning | varies |
| Day-to-day biological variation in the same man | larger than assay error |
| Which machine ran the sample | worst at low values |
Start with the most striking one. In 2013, researchers gave 74 men a standard glucose drink — the sugary one used for diabetes testing — and measured testosterone repeatedly for two hours. Mean testosterone fell 25%, and was still down at the two-hour mark. [2]
Then the part that matters for screening. Of the 66 men who had normal testosterone before the drink, ten — 15% — fell into the hypogonadal range at one or more points afterwards. [2]
One man in seven can be converted from normal to clinically low by a glass of glucose. Nothing changed about his body. Something changed about when you took the blood.
This is precisely why the guideline specifies a fasting morning sample, and then says to confirm it with a second one. [1] In its own words, the aim is to avoid labelling men based on “transient medical disorders, biological variations in T concentrations, technical variations and inaccuracy in T measurements, or SHBG abnormalities.” [1]
Four separate ways to get a wrong answer, listed by the people who wrote the standard.
The number moves even when you do everything right
A study of 132 men aged 30 to 79 took repeated blood samples — pairs 1 to 3 days apart, then again at three months and six months, up to six samples each. Its conclusion: “Biological variation generally accounted for more of total intraindividual variation than did assay variation.” [3]
Translated: your testosterone genuinely bounces around on its own, and it bounces around more than the machine’s error does. Two readings from the same healthy man on two mornings are not the same reading.
And the machine has its own problem, in the worst possible place. A review of testosterone assay performance found that “within-assay variability for current assays is generally high, especially at low analyte concentrations.” [4]
Read that twice. The test is least reliable at low values — which is the only range where the result changes what happens to you. Standardisation has improved: a US Centers for Disease Control programme cut mean absolute bias against the reference method by half between 2007 and 2011. [4] But “better” is not “interchangeable”.
What counts as normal, and why your lab’s range might not
Here is the good news, and it is genuinely useful. In 2017 researchers took 9,054 men from four large cohort studies in the US and Europe, cross-calibrated the assays against a CDC reference method, and produced a harmonised range. [5]
For healthy non-obese men aged 19 to 39, the middle 95% runs from 264 to 916 ng/dL, with a median of 531. [5]
Two things follow. First, that range is enormous — the man at the top has roughly three and a half times the testosterone of the man at the bottom, and both are normal. Second, and more awkward: the authors found that “a substantial proportion of intercohort variation in testosterone levels is due to assay differences.” [5]
Some of the apparent difference between groups of men was never about the men. It was about the machines. If your result comes back near a cutoff, the laboratory that ran it is part of the answer.
Who should get tested, then?
The guideline does not say nobody. It gives a list — groups where low testosterone is common enough and treatment plausible enough that measuring makes sense: men with low libido, erectile dysfunction, infertility, HIV-associated weight loss, osteoporosis or a low-trauma fracture, a history of anabolic steroid use, and men taking opioids or other drugs that affect testosterone. [1]
Being over 30 is not on that list.
This is the distinction the whole argument turns on, and it is not a fringe position. A 2019 review compared the guidelines of seven international bodies — urological, endocrine and sexual-medicine societies across the US and Europe — and found them in agreement that clinicians should “consider screening asymptomatic men with certain conditions that increase the risk”. [6]
Conditions. Not ages. Every one of those bodies endorses targeted testing; none endorses testing a population because it had a birthday.
What about the men who do turn out to be low?
If you have symptoms and two confirmed low morning readings, testosterone therapy is a real option with real evidence, and one large question about it has been answered well.
TRAVERSE, published in 2023, randomised 5,246 men aged 45 to 80 who already had cardiovascular disease or were at high risk of it, all of whom had symptoms and two fasting testosterone readings under 300 ng/dL. Over a mean 33 months of follow-up, major cardiovascular events occurred in 7.0% on testosterone versus 7.3% on placebo — a hazard ratio of 0.96, meeting the trial’s noninferiority bar. [7]
That is a genuinely reassuring result on the question people were most worried about. It is worth noticing what it is not. TRAVERSE enrolled men the guideline pathway had already identified — symptoms first, then two confirmed low readings. It tells you that treating those men does not appear to raise cardiovascular risk. It does not tell you that finding more of them by screening would do anybody good, because that trial has never been run.
And there is a cost that does not show up in a cardiovascular endpoint. All seven guideline bodies agree on it: taking testosterone impairs sperm production. [6] For a 30-year-old who has not finished having children, that is not a footnote.
The verdict
Testing every man over 30 is not supported by the evidence, and the body that writes the standard recommends against it in one sentence.
Not because testosterone does not matter, and not because low testosterone is not real — it is, it has consequences, and men who have symptoms should absolutely get checked. But a screening programme makes a specific promise: that finding the thing early, in people who have not complained of it, leads to better outcomes. For testosterone, nobody has demonstrated that. No trial has even tried.
What screening reliably produces instead is a population of men holding a number they now have to do something about — a number that a breakfast can move by a quarter, that varies between two ordinary mornings, and that the machine measures least accurately in exactly the range where it decides.
The guideline says this outright. It “places a high value on avoiding labeling, testing, treating, and monitoring healthy men for whom the benefits and risks are unclear.” [1]
If you are getting tested anyway — because a policy requires it, or because you want to — the practical advice is the same advice the guideline gives a doctor. Go in the morning. Go fasted. Do not accept a diagnosis from a single draw; the standard is two. Ask which assay the lab used. And treat one number as what it is: a snapshot of a moving thing, not a verdict on you.
The most useful question is not “what is my number”. It is “do I have symptoms” — because that is the question the guideline asks first, and the number only means something after it has been answered.
This is the first in a series on testosterone. Still to come: what a normal level actually is, free versus total, whether boosters do anything, and what the training and sleep evidence really shows.
Related: ashwagandha benefits · berberine benefits · does lack of sleep cause weight gain · is creatine safe
