Testosterone does look lower in each generation. The assays make it hard to be sure.
Two datasets report an age-independent fall in men's testosterone across birth cohorts. A separate study of nine thousand men found much of the variation between cohorts came from the tests themselves.
The claim that men's testosterone is lower with each passing generation rests on two real datasets, and both of them report the same thing: a fall over calendar time that is not explained by the men in the sample getting older. What neither can rule out cleanly is the possibility that a meaningful share of the difference is in the laboratory rather than in the men — because a separate study of more than nine thousand men found that assay differences accounted for a substantial part of the variation between cohorts measuring the same hormone.
The original finding
The Massachusetts Male Aging Study followed randomly selected men aged 45 to 79 in greater Boston across three data collection waves: a baseline running from 1987 to 1999, and follow-ups in 1995–1997 and 2002–2004 [s1]. It produced 2,769 observations on 1,532 men — 1,374 at the first wave, 906 at the second and 489 at the third [s1].
The 2007 analysis in the Journal of Clinical Endocrinology and Metabolism reported a substantial age-independent decline in total and calculated bioavailable testosterone, one the authors described as not attributable to observed changes in explanatory factors including smoking and obesity [s1]. The population-level declines were larger than the cross-sectional declines usually attributed to ageing [s1]. The paper's own conclusion attributes the fall to birth cohort differences, or to health or environmental effects not captured in the data — which is a statement about what the study could not measure, not a mechanism [s1].
The replication in younger men
The obvious criticism of a Boston cohort of middle-aged and older men is that it says nothing about young men. A 2020 analysis in European Urology Focus addressed that using the National Health and Nutrition Examination Surveys, covering 4,045 men from 1999 to 2016 [s2]. After controlling for confounders, mean total testosterone was lower in the later survey cycles (2011–2016) than in the earliest (1999–2000), with all comparisons at p < 0.001 [s2].
Higher body mass index was associated with lower total testosterone, as expected — but the downward trend over time remained significant even among men with normal BMI [s2]. That is the finding that makes the story more than a restatement of rising obesity. The authors list the limitations themselves, and one of them is the point this article turns on: the surveys used differing assays to measure total testosterone across the period [s2].
Why the assay caveat is not a technicality
In 2017, a group including researchers from the CDC set out to build reference ranges for testosterone that would hold across laboratories, using 9,054 community-dwelling men from four cohort studies in the United States and Europe: the Framingham Heart Study, the European Male Aging Study, the Osteoporotic Fractures in Men Study, and the Male Sibling Study of Osteoporosis [s3]. They measured testosterone in 100 participants from each cohort using a reference method at the CDC, then derived equations to convert each cohort's values onto that common scale [s3].
Harmonisation reduced the variation between cohorts in men of similar ages, and the paper's conclusion states plainly that a substantial proportion of the intercohort variation in testosterone levels is due to assay differences [s3]. Once harmonised, age-specific concentrations in non-obese men were similar across all four cohorts [s3].
That is a direct problem for any secular-trend claim built by comparing measurements taken years apart on different platforms. If four contemporaneous cohorts disagreed with each other largely because of how they measured, then two survey cycles fifteen years apart, using different assays, can differ for the same reason.
What a normal value even is
The same harmonisation study produced the numbers that now anchor much of the clinical conversation. In healthy non-obese men aged 19 to 39, the harmonised 2.5th, 5th, 50th, 95th and 97.5th percentiles were 264, 303, 531, 852 and 916 ng/dL respectively — giving a normal range of 264 to 916 ng/dL for that group [s3]. The spread is the point: the fifth percentile and the ninety-fifth differ by a factor of nearly three among men who are healthy by definition.
A population whose average moves within a range that wide is not obviously a population that has crossed a clinical threshold. It is a population whose average moved.
Where this leaves the question
The generational-decline claim is better supported than most viral health claims and weaker than it is usually presented. Two independent datasets, in different age bands and different decades, both report an age-independent fall [s1] [s2]. Neither was designed to separate a real biological shift from a measurement shift, and the one study that directly tested how much measurement contributes found the contribution substantial [s3].
There is also a gap nobody has filled. None of these analyses establishes what a population-level change in average testosterone means for any individual man's health, because the outcome measured was the hormone concentration, not anything that happened to the men. That question — whether the trend, if real, matters — remains open, and the honest reporting of it is that the trend has been measured far more often than its consequences have.
Sources
- A population-level decline in serum testosterone levels in American men — Journal of Clinical Endocrinology and Metabolism, 2006-10-24
- Decline in Serum Testosterone Levels Among Adolescent and Young Adult Men in the USA — European Urology Focus, 2020-02-18
- Harmonized Reference Ranges for Circulating Testosterone Levels in Men of Four Cohort Studies in the United States and Europe — Journal of Clinical Endocrinology and Metabolism, 2017-01-10
Sources
- A population-level decline in serum testosterone levels in American men — Journal of Clinical Endocrinology and Metabolism , October 24, 2006
- Decline in Serum Testosterone Levels Among Adolescent and Young Adult Men in the USA — European Urology Focus , February 18, 2020
- Harmonized Reference Ranges for Circulating Testosterone Levels in Men of Four Cohort Studies in the United States and Europe — Journal of Clinical Endocrinology and Metabolism , January 10, 2017
More on
A review of 253 testosterone clinic websites found a quarter breaking guidelines
Researchers searched in four languages across three engines. One in ten clinics said it would prescribe testosterone to men with normal levels; one in five claimed it cuts cardiovascular risk.
Testosterone therapy improved sexual function, not energy. Then it broke more bones.
The randomised evidence in older men found a moderate effect on sexual function and none on vitality or walking. A 5,246-man safety trial cleared it on heart attacks and found more fractures.
Placebo responses in US pain trials grew for two decades. Drug responses did not.
Analgesic trials keep failing, and one reason is that the comparison arm keeps improving. Where a trial recruits, how long it runs and how big it is all move the placebo response — and none of that is about the drug.
95% of human-microbiome-into-mice studies reported the disease transferred too
A systematic review in Cell called that rate implausible and said it overstates the gut microbiome's role in human disease. A separate audit found the statistical tools themselves disagree.