Low Testosterone Statistics: Why Estimates Range From 2.1% to 38.7%

Q: Why do low testosterone statistics vary so much?

The studies do not count the same thing. A laboratory cutoff produces a much larger number than a rule that also requires low free testosterone and specific symptoms.

By TRT Provider Guide
Last verified: September 2, 2026

Low testosterone statistics can look as if they disagree. One study says 2.1%. Another says 38.7%. Both numbers were published. They do not measure the same group of men or use the same rules.

The number to quote: Across the four primary studies compared on this page, estimates tied to “low testosterone” run from 2.1% to 38.7%—an 18.4-fold spread. The studies used different ages, settings, blood-test rules, and symptom requirements, so the figures are not interchangeable.

The cleanest same-sample comparison: In the BACH community study, approximately 24% of men had total testosterone below 300 ng/dL. Only 5.6% met the study’s full rule of low total testosterone, low free testosterone, and its specified symptom rule. Using the paper’s rounded 24% figure, the full definition counted about 77% fewer men.

The first quote is the broad cross-study finding. The second is the better example of what a stricter definition can do inside one group of men.

Scope note: These are study estimates, not a way to diagnose one person. Major medical guidelines require symptoms or signs plus repeat morning blood testing before a diagnosis is made.

Bar chart showing that approximately 24% of the same BACH sample had total testosterone below 300 ng/dL, while 5.6% met the full low-total, low-free, and symptom definition.

Figure 1. The definition changed the estimate in the same 1,475 men. This is a classification comparison, not a treatment effect. The paper reported the laboratory-only value as approximately 24%, so the calculated reduction is also approximate.

Source: Araujo et al., Prevalence of Symptomatic Androgen Deficiency in Men. Download the figure as SVG or view the chart data.

Low testosterone statistics at a glance

38.7%: In the HIM study, men age 45 or older seen at 130 U.S. primary-care practices were classified by total testosterone below 300 ng/dL or current androgen treatment. Symptoms were not required.
25.7%: In a 2026 NHANES analysis, 10,357 U.S. men age 18 or older were classified by total testosterone below 300 ng/dL. Testosterone users were excluded, and symptoms were not required.
5.6%: Men age 30 to 79 in the Boston Area Community Health study who met its combined low-total-testosterone, low-free-testosterone, and specified symptom rule.
2.1%: Sixty-three of 2,966 men with complete data in the European Male Ageing Study met its late-onset hypogonadism rule: three sexual symptoms plus low total and free testosterone.
18.4×: The arithmetic spread between 38.7% and 2.1%.
10,398 → 10,357: The 2026 NHANES paper identified 41 exogenous testosterone users among men with valid laboratory and survey-design data, then excluded them. The 41 is an unweighted count; the paper’s 0.1% is survey-weighted.

The year 2026 in the title is the verification year. It is not the year every study collected its data.

Low testosterone prevalence data by study

Table 1. Four published estimates that are often treated as if they answer the same question
Primary study	Population and setting	Rule used by the study	Published estimate	What the estimate can support
HIM study, Mulligan et al. (2006)	2,162 men age 45 or older seen at 130 U.S. primary-care practices	Total testosterone below 300 ng/dL or current androgen treatment; no symptom rule	38.7%	A biochemical/treatment classification in an older clinic sample. It is not a national diagnosis rate.
Goulian et al. / NHANES (2026)	10,357 U.S. men age 18 or older across the 2011–2016 and 2021–2023 cycles	Total testosterone below 300 ng/dL; exogenous testosterone users excluded; no symptom rule	25.7%	A survey-weighted biochemical estimate, not a symptom-confirmed diagnosis rate.
BACH, Araujo et al. (2007)	1,475 men age 30 to 79 in a Boston community sample	Total testosterone below 300 ng/dL and free testosterone below 5 ng/dL and the study’s specified symptom rule	5.6%	Symptomatic androgen deficiency under the study’s full rule.
EMAS, Wu et al. (2010)	2,966 men with complete data, age 40 to 79, at eight European centers; 3,369 entered the initial survey	Three sexual symptoms and total testosterone below 11 nmol/L and free testosterone below 220 pmol/L	2.1%	Late-onset hypogonadism under the EMAS research definition. It is not a U.S. estimate.

Source: Mulligan et al., HIM study; Goulian et al., NHANES analysis; Araujo et al., BACH; Wu et al., EMAS. Calculation: 38.7 ÷ 2.1 = 18.4286, rounded to 18.4.

Horizontal bar chart comparing four published low-testosterone estimates: 38.7% in HIM, 25.7% in a 2026 NHANES analysis, 5.6% in BACH, and 2.1% in EMAS. The studies used different populations and definitions.

Figure 2. Four published estimates answer four different questions. This is a descriptive comparison, not a pooled estimate. The studies differ by age, setting, country, test rule, and symptom rule.

Source: The four primary studies in Table 1. Download the figure as SVG or view the chart data.

What the 18.4-fold number does—and does not—mean

The 18.4-fold spread is real arithmetic. It is not a pooled prevalence estimate. It does not prove that one definition alone caused the whole gap. The studies also used different ages, countries, and settings.

That is why the BACH comparison matters. It holds the sample still. In the same group, approximately 24% met the total-testosterone cutoff, while 5.6% met the full low-total, low-free, and symptom definition.

Table 2. The BACH definition ladder in the same community sample
Step	What was required	Estimate
Blood result only	Total testosterone below 300 ng/dL	Approximately 24%
Full BACH definition	Low total testosterone, low free testosterone, and the study’s specified symptom rule	5.6%
Approximate difference	24.0 minus 5.6 percentage points	About 18.4 points
Approximate relative reduction	(24.0 − 5.6) ÷ 24.0	About 76.7%, or 77%
Approximate ratio	24.0 ÷ 5.6	About 4.3×

The BACH symptom rule was low libido, erectile dysfunction, osteoporosis or fracture, or at least two of these four nonspecific symptoms: sleep disturbance, depressed mood, lethargy, or diminished physical performance.

Source: Araujo et al., Prevalence of Symptomatic Androgen Deficiency in Men. The paper reported the laboratory-only result as approximately 24%, so every calculation based on that rounded value is approximate.

Why one low testosterone test is not a diagnosis

A blood-test cutoff and a clinical diagnosis answer different questions.

The American Urological Association testosterone deficiency guideline says clinicians should use total testosterone below 300 ng/dL as a reasonable cutoff. It also says the diagnosis should be made only after two total-testosterone measurements on separate days, both taken in the early morning, and only when low testosterone is joined by symptoms or signs.

The Endocrine Society guideline also requires symptoms or signs plus testosterone levels that are clearly and consistently low. It calls for a repeat morning fasting total-testosterone test to confirm the result.

Table 3. Research cutoff versus clinical diagnosis
Question	What is needed
“Was this blood result below the study cutoff?”	One study-defined laboratory threshold may be enough for that research count.
“Does this man meet the AUA diagnostic rule?”	Symptoms or signs, plus two separate early-morning total-testosterone results, using below 300 ng/dL as a reasonable cutoff.
“Does this man meet the Endocrine Society diagnostic rule?”	Symptoms or signs, accurate testing, and clearly and consistently low testosterone confirmed with a repeat morning fasting test.
“Does this man meet the EMAS research definition?”	Three named sexual symptoms plus low total and free testosterone under the EMAS thresholds.

Source: American Urological Association, Testosterone Deficiency Guideline; Endocrine Society clinical practice guideline; Wu et al., EMAS.

U.S. low testosterone data and European data are not interchangeable

The 38.7% HIM estimate came from U.S. primary-care patients age 45 or older. The 5.6% BACH estimate came from a Boston community sample age 30 to 79. The 2.1% EMAS estimate came from men age 40 to 79 at eight European centers.

Those differences matter before the laboratory rules are even considered. Older clinic patients are not the same population as community-dwelling men across a wider age range. A European research definition is not a direct estimate for all U.S. men.

The safe wording is simple:

“Published estimates vary from 2.1% to 38.7% across selected studies with different populations and definitions.”

The unsafe wording is:

“Between 2.1% and 38.7% of all men have low testosterone.”

The first statement describes the evidence. The second pretends the studies measured one universal population in one universal way.

What the 41 excluded testosterone users mean

The 2026 NHANES analysis started with 12,357 eligible men. It excluded 1,959 with missing testosterone results or incomplete survey-design data, leaving 10,398. The authors then identified 41 men reporting exogenous testosterone therapy and excluded them, leaving a final prevalence sample of 10,357.

The results section reports those users as 0.1%, n = 41. The 41 is the observed count. The 0.1% is survey-weighted. Raw division gives about 0.39%, but a raw sample fraction is not expected to match a weighted national estimate.

The clean wording is:

“The study identified 41 exogenous testosterone users in the unweighted sample, reported a survey-weighted share of 0.1%, and excluded those users before estimating prevalence in 10,357 men.”

Source: Goulian et al., Unmasking Masculinity: Biological, Behavioral, and Social Determinants of Testosterone Deficiency in U.S. Men, 2011–2023.

What changed in FDA testosterone labeling in 2025 and 2026

On February 28, 2025, the U.S. Food and Drug Administration announced class-wide labeling changes for prescription testosterone products. FDA said manufacturers should remove boxed-warning language about increased cardiovascular risk after its review of the TRAVERSE trial. It also required warnings about increased blood pressure based on ambulatory blood-pressure studies. The 2025 action retained the limitation-of-use language for age-related hypogonadism.

On June 18, 2026, HHS announced that FDA was requesting another round of label updates. The requested changes would remove that limitation-of-use statement, update prostate-cancer information, and revise warnings about benign prostatic hyperplasia. That announcement described requested label changes; it should not be rewritten as proof that every product label had already been updated.

Neither announcement created a new prevalence estimate or turned one low blood result into a diagnosis.

Source: FDA, “FDA issues class-wide labeling changes for testosterone products,” February 28, 2025; HHS, “HHS Announces Requested Updates to Testosterone Therapy Product Labels,” June 18, 2026; Lincoff et al., TRAVERSE.

Which low testosterone statistic should a writer use?

Use the number that matches the sentence you are writing.

Table 4. Best number by claim
Claim you need to support	Best number	Required wording
Show how wide published estimates can be	2.1% to 38.7%; 18.4× spread	Say the studies used different populations and definitions.
Show what a stricter rule did in one sample	Approximately 24% to 5.6%; about 77% fewer	Say BACH added both low free testosterone and its specified symptom rule to the low-total-testosterone rule.
Describe the HIM clinic sample	38.7%	Say men were age 45 or older and seen in U.S. primary-care practices; do not call it all U.S. men.
Describe symptomatic androgen deficiency in BACH	5.6%	Use the full low-total, low-free, and symptom definition.
Describe the EMAS late-onset hypogonadism definition	2.1%	Say it was a European study and name the symptom-plus-laboratory rule.
Describe the exclusions in the 2026 NHANES analysis	41 unweighted users; 0.1% weighted estimate; 10,357 final sample	Say 41 users were removed from 10,398 men with valid laboratory and design data. Do not treat the raw fraction as the weighted estimate.

Source: Primary studies listed in Tables 1–3 and the 2026 NHANES sample-flow section above.

Methodology

Question this page answers

How much do published low-testosterone estimates change when studies use different populations, laboratory cutoffs, and symptom rules?

Study selection

This comparison keeps four primary studies that publish a clear estimate and state the population and rule behind it. It includes broad biochemical estimates and stricter symptom-plus-biochemical estimates because the point is to show what each number actually counts.

Secondary summaries, commercial surveys, unsourced percentages, and claims with no usable denominator were not used for the headline comparison.

Data extraction

For each study, the page records:

the population and setting;
the age range;
the total-testosterone rule;
any free-testosterone rule;
any symptom rule;
the published estimate; and
the main limit on interpretation.

Published percentages are kept as reported. No pooled prevalence estimate was created.

Calculations

Cross-study spread: 38.7 ÷ 2.1 = 18.428571, reported as 18.4×.
Approximate BACH percentage-point difference: 24.0 − 5.6 = 18.4 percentage points.
Approximate BACH relative reduction: (24.0 − 5.6) ÷ 24.0 × 100 = 76.7%, rounded to about 77%.
Approximate BACH ratio: 24.0 ÷ 5.6 = 4.3×.

The BACH paper says “approximately 24%.” The calculations above use 24.0 only to make the arithmetic reproducible; they remain approximate.

What was not done

This is not a systematic review or meta-analysis. The studies were not pooled because their populations and definitions are too different for one combined percentage to be honest. The page does not estimate the number of men in 2026 who have a clinical diagnosis.

Verification record

All headline values, study definitions, sample descriptions, calculations, and current guideline or FDA statements were checked against the linked primary paper, issuing medical society, or federal agency on September 2, 2026.

Data limitations

The four percentages are not four measurements of one identical population.
The HIM sample came from primary-care practices and included only men age 45 or older.
BACH and EMAS used symptom rules that the broad biochemical estimates did not use.
A low value in a research dataset is not automatically a clinical diagnosis.
Testosterone assays, timing rules, and thresholds differ across studies.
In the 2026 NHANES paper, 0.1% is survey-weighted; 41 is an unweighted observed count; and 10,357 is the final sample after those users were excluded.
The data do not tell a reader whether treatment is right for one person.

Related research

How to cite this page

This block gives neutral attribution details for the page and its original cross-study calculations.

Suggested citation: TRT Provider Guide. “Low Testosterone Statistics: Why Estimates Range From 2.1% to 38.7%.” TRT Provider Guide. Published 2026-09-02. Last verified September 2, 2026. https://trtproviderguide.com/research/low-testosterone-statistics/

Dataset title: Low testosterone prevalence definition comparison
Version: 2026-09-02
Publisher: TRT Provider Guide
Method: Manual extraction from the cited primary studies; calculations shown above; no pooled meta-analysis.

Frequently asked questions

What percentage of men have low testosterone?

There is no single honest percentage. In the four primary studies compared here, published estimates ran from 2.1% to 38.7%. The low and high figures used different ages, settings, tests, and symptom rules.

Why do low testosterone statistics vary so much?

The studies do not count the same thing. A single total-testosterone result below a cutoff produces a much larger number than a rule that also requires low free testosterone and specific symptoms.

Does one testosterone result below 300 ng/dL mean a man has hypogonadism?

No. The American Urological Association uses below 300 ng/dL as a reasonable cutoff, but it also requires symptoms or signs and two early-morning total-testosterone tests on separate days. The Endocrine Society also calls for symptoms plus consistently low levels confirmed with a repeat morning fasting test.

Is 300 ng/dL the only accepted cutoff?

No. The AUA uses below 300 ng/dL as a reasonable cutoff. The Endocrine Society tells clinicians to use accurate assays and a well-derived lower limit for the assay or reference range. Research studies also use other total- and free-testosterone thresholds.

What is the cleanest statistic to quote?

For an across-study summary, use the 2.1% to 38.7% range and say it is an 18.4-fold spread among studies with different populations and definitions. For a same-sample example, use BACH: approximately 24% met the total-testosterone cutoff, while 5.6% met the full low-total, low-free, and symptom definition—about 77% fewer.

How did the study report 41 testosterone users as 0.1%?

The 2026 NHANES analysis had 10,398 men with valid laboratory and survey-design data, identified 41 reporting exogenous testosterone, and excluded them to leave 10,357. The 41 is an unweighted count. The 0.1% figure is survey-weighted. Those two measures should be labeled separately.