Are men really collapsing? What the data says about testosterone and sperm decline.

You have probably seen the headlines. Testosterone is down by a third. Sperm counts have halved. Modern men are weaker, less fertile, and biologically diminished compared with their fathers and grandfathers. The story is everywhere — podcasts, supplement ads, clinic marketing, political speeches, and social media threads about microplastics, seed oils, smartphones, and the supposed collapse of masculinity. It is one of the most alarmist health narratives of the past decade. It is also much less settled than it sounds. This article walks through what the actual studies say, where the famous numbers come from, and what the honest answer is when you look at the cleanest data we have.

The testosterone story: what the famous 25% drop actually shows

The most-cited testosterone claim is that levels in young American men fell by around 25% between 1999 and 2016. That figure comes from a 2021 study by Lokeshwar and colleagues using NHANES, the large US national health survey. The reported numbers were striking: average testosterone in 15 to 39 year old men fell from 605 ng/dL in 1999–2000 to 451 ng/dL in 2015–2016. On the surface, that sounds like biological collapse.

But there is a major catch: testosterone measurement itself changed during that period. Older datasets relied on antibody-based immunoassays, which are less specific than modern methods and can over-read testosterone because antibodies may cross-react with other steroid molecules. From 2011 onwards, NHANES moved into the modern mass spectrometry era — methods that separate and identify testosterone far more accurately. If the ruler changes, the trend can look bigger than the biology.

The threshold did not change. The 300 ng/dL cutoff remained widely used, even though modern reference data from the Endocrine Society's four-cohort harmonisation study suggest the lower limit in healthy young men is closer to 264 ng/dL. That does not make 300 meaningless, but it does mean using it as a hard single-draw trigger can overcall low testosterone.

Now look at what happened once measurement became consistent. A 2025 analysis of NHANES data from 2011 to 2023, covering more than 10,000 men, found that testosterone deficiency prevalence fell from 28.1% in 2011 to 20.3% in 2021–2023. Age-standardised, prevalence dropped from 29.3% to 22.4%, a statistically significant trend. In other words: within the modern measurement era, US testosterone deficiency appears to have fallen, not risen.

That does not mean nothing happened. The Massachusetts Male Aging Study, one of the cleaner long-term studies, used the same blood test, the same lab, and the same technician across three waves between 1987 and 2004. It found a real age-matched decline of about 1.2% per year. So the honest testosterone story is not "nothing happened" — it is that a real, modest decline probably occurred in some populations between the 1980s and early 2000s, but the viral 25–30% collapse story is exaggerated by assay changes and outdated thresholds.

The sperm story: a 50% decline that depends on which study you read

The sperm story is even more famous. The headline claim is that sperm counts have fallen by roughly 50% since the early 1970s. This comes mainly from a series of meta-analyses by Hagai Levine and colleagues. Their 2017 paper pooled data from 42,935 men across 185 studies between 1973 and 2011 and reported that sperm concentration in unselected Western men fell from 99 million per mL in 1973 to 47 million per mL in 2011 — a 52% decline. Their 2023 update expanded the dataset globally to more than 57,000 men and again reported a decline, claiming it had accelerated after 2000.

This is the foundation of much of the "spermageddon" discourse. But it is also heavily contested.

A 2023 meta-analysis by Cipriani and colleagues looked only at high-income countries and post-2000 data, covering 24,196 men. It found no statistically significant change in sperm concentration or total sperm count. None of the analyses across young unselected men, sperm donors, or fertile populations reached statistical significance.

A 2025 US-only systematic review by Lewis and colleagues, published in Fertility and Sterility, looked at 11,787 American men across 58 studies between 1970 and 2018. It found no significant decline in sperm concentration among fertile or unselected US men. The slope was actually slightly positive. This is the cleanest single-country recent analysis available, and it substantially undermines the alarmist narrative for the United States specifically.

Then there is the Danish military draftee cohort — one of the cleanest population-based datasets we have. It included 4,867 19-year-old men sampled with consistent methodology between 1996 and 2010. Median sperm concentration rose from 43 to 48 million per mL, and median total sperm count rose from 132 to 151 million.

So depending on which dataset you look at, sperm counts have halved, stayed the same, or risen modestly. That is not a trivial disagreement — it tells us the answer depends heavily on study design. The Levine meta-analyses pool together very different populations: fertile men, unselected men, sperm donors, men from fertility clinics, different countries, different decades, and different laboratory methods. That does not make the findings useless. But it does make the headline number much less clean than it is often presented. The strongest conclusion is not that sperm decline is fake. It is that the global sperm decline narrative is much more methodologically fragile than the headlines suggest.

Does lower sperm count actually mean lower fertility?

This is where the story gets even more important. Sperm concentration matters — very low sperm counts clearly reduce fertility. But sperm count is not fertility. Natural conception depends on sperm concentration, total count, motility, morphology, DNA fragmentation, female age, ovulation timing, intercourse frequency, reproductive anatomy, and the length of time a couple has been trying.

There also appears to be a threshold effect. A landmark study by Bonde and colleagues, looking at 430 Danish first-pregnancy planning couples, found that the probability of conception increased with sperm concentration up to about 40 million per mL. Above that threshold, additional sperm did not translate into additional fertility. So even if a pooled average fell from 99 million per mL to 47 million per mL, that does not necessarily mean population-level fertility has halved — both values are above the most important fertility threshold, and the relationship between sperm count and pregnancy probability is not linear.

There is no clear evidence that declining birth rates in developed countries are primarily being driven by a biological male fertility collapse. That does not mean male fertility is irrelevant. It means sperm-count headlines should not be confused with population birth-rate statistics.

What's actually driving the real decline — and it's not microplastics

The biggest measurable driver of the decline that is real is not microplastics, seed oils, or smartphones. It is obesity. The link between body fat and male hormonal health is one of the most consistent findings in the entire literature.

Fat tissue is hormonally active. It contains aromatase, an enzyme that converts testosterone into oestrogen. The more excess fat tissue a man carries, the more testosterone gets converted away from the androgen pathway. Obesity also drives chronic inflammation, insulin resistance, sleep apnoea, lower SHBG (a protein that carries testosterone in the blood), impaired signalling from the brain to the testes, and poorer testicular function. These pathways affect both testosterone production and sperm production. A meta-analysis by Sermondade and colleagues, including more than 13,000 men, found that obese men had a 28% higher rate of abnormally low sperm counts than lean men, while morbidly obese men had more than double the rate.

The cleanest natural experiment we have

The most important finding on the obesity question comes from Denmark. In 2007, Andersson and colleagues published a study analysing four large Danish population surveys covering more than 5,000 men aged 30 to 70. The methodology was exceptionally clean: stored serum samples from across two decades were thawed and analysed together using the same assay platform in the same run, eliminating the methodology-drift problem that plagues most secular decline studies.

The raw finding was a secular decline in total testosterone across the period. The Danish data looked just like the American data — men's testosterone appeared to be falling generation by generation. But the authors did something most studies skip: they adjusted for the concurrent rise in BMI. The decline disappeared. After controlling for BMI, the secular trend in total testosterone was no longer statistically significant. Obesity explained essentially all of it.

This is the cleanest natural experiment on the question that exists in the literature. Same population, same lab, same assay, samples measured together, same statistical model. Adjust for the obesity epidemic and the apparent testosterone crisis simply is not there. This is the part of the conversation the panic narrative skips entirely. We've explored this connection in more depth in our piece on the real health gap between men and women — much of what looks like a mystery decline is, in plain terms, a metabolic problem.

Two behavioural confounders nobody talks about

There are two behavioural variables that have changed substantially over the same period as the alleged biological decline, but are barely captured in the major decline studies: sleep and ejaculation frequency.

Sleep

Sleep is not a minor lifestyle detail — it is part of male endocrine regulation. In a controlled JAMA study, healthy young men restricted to five hours of sleep per night for one week had a 10–15% reduction in daytime testosterone, an effect comparable to 10–15 years of aging. Population sleep data is striking: average US adult sleep duration has fallen from around 8 hours in the 1960s to around 6.8 hours today, and the proportion of US adults reporting less than 7 hours of sleep has risen from around 22% in 1985 to 35% in 2014.

Yet most long-term testosterone decline studies controlled for age, BMI, smoking, alcohol, and comorbidities — while sleep duration and sleep quality were either not measured or not modelled in a way that can explain secular trends. If men are sleeping less, sleeping worse, or developing more sleep apnoea over time, some of the apparent hormonal decline reflects sleep-related endocrine suppression rather than mysterious biological deterioration. We've written more about why sleep is the foundation of men's health.

Ejaculation frequency

Semen analysis is highly sensitive to abstinence time. Labs require 2 to 7 days of abstinence before producing a sample because the effect on measured sperm parameters is large — a man's sperm concentration can roughly double between 1 day and 5 days of abstinence. Major studies adjust for the abstinence period immediately before sample collection, but that is not the same as measuring a man's usual ejaculation frequency over the preceding weeks and months.

UK survey data found that the proportion of men aged 16 to 44 reporting masturbation in the past month rose from 73% in 1990 to 95% in 2010. Similar patterns appear in US data. The most likely drivers are reduced stigma, broader cultural acceptance, and universal availability of high-speed internet pornography from roughly 2000 onwards. If habitual ejaculation frequency has risen across generations, some part of the difference in measured semen parameters reflects more frequent emptying of sperm reserves rather than reduced testicular production. The major decline meta-analyses do not control for this, because the data simply is not available.

The diagnosis problem: why one test is not a diagnosis

There is one more issue, because it shapes how these population findings get translated into individual clinical decisions. Both testosterone testing and semen analysis suffer from the same underlying problem: they are based on single measurements of biological variables that swing widely from day to day.

Why a single low testosterone result is not a diagnosis

Testosterone fluctuates by time of day and from day to day. Levels are highest in the morning and fall through the day. Sleep, stress, illness, calorie restriction, alcohol, exercise, and lab variation can all move the number. Day-to-day variability is wide enough that a man whose true average is around 350 ng/dL can test at 270 one Monday and 410 the next — same lab, no intervention.

Serious clinical guidelines from the Endocrine Society, the American Urological Association, and the European Association of Urology all require two confirmed morning blood draws on separate days before diagnosing testosterone deficiency, plus matching symptoms. The European Male Aging Study, which used proper repeat testing and structured symptom assessment, found that the prevalence of clinically meaningful late-onset hypogonadism in men aged 40 to 79 was only 2.1% — far below the 20–30% headline figures circulating in popular media.

Yet much of the low-testosterone industry collapses all of that nuance into one message: your number is under 300, you have low T, you need treatment. That is not good medicine — it is a sales funnel. True hypogonadism exists, and for men who have it, testosterone therapy can be genuinely life-changing. But many men being told they have low T have something else: reversible metabolic suppression caused by poor sleep, excess body fat, alcohol, medications, stress, or sleep apnoea. For those men, the first-line intervention is not testosterone — it is fixing the physiology that is suppressing it. This is also why most testosterone boosters don't work: they are aimed at the wrong target.

Why a single semen analysis is not a diagnosis either

The same brittleness applies to semen analysis. Sperm concentration, motility, and morphology vary substantially from sample to sample in the same man, depending on abstinence time, recent illness, stress, alcohol, heat exposure, and even the conditions under which the sample was produced. Within-individual variation in sperm concentration can exceed 30% from one sample to the next. This is why the World Health Organization explicitly recommends at least two semen analyses, ideally three, separated by several weeks, before making any diagnostic conclusions. In real-world practice, most men receive a single sample analysis and a verdict based on that one number.

What we can honestly say — and what to do about it

The full picture across the testosterone and sperm literature is more nuanced than either side wants it to be. The alarmist version says male biology is collapsing. The dismissive version says nothing is happening. Both are too simple.

A more accurate summary:

• There has probably been a real, modest decline in testosterone and sperm parameters in some male populations between roughly the 1980s and mid-2000s.
• The decline is likely much smaller than the famous headline numbers suggest.
• The 25–30% testosterone collapse is partly distorted by assay changes, outdated thresholds, and incomplete adjustment for behavioural factors. In the Danish MONICA data, adjusting for obesity made the decline statistically disappear.
• The 50% sperm-count collapse is heavily dependent on pooled meta-analyses that combine very different populations and methods. The cleanest single-country recent meta-analysis found no significant decline in US men. The cleanest population-based cohort, of Danish 19-year-olds, found a slight rise.
• In cleaner modern datasets, the picture is mixed and often more reassuring. US testosterone deficiency has fallen since 2011.
• The strongest common driver of the real decline is metabolic health — obesity, poor sleep, physical inactivity, alcohol, insulin resistance, and chronic disease all suppress male reproductive physiology.

So are men really collapsing? By the cleanest available data, no. There was probably a real, modest, mostly metabolic decline in some reproductive markers through the late 20th century and early 2000s. But within the modern measurement era, the picture has stabilised, become mixed, or even improved in some datasets.

What is true is that modern men are metabolically less healthy than they should be. They are heavier. They move less. Many sleep badly. Many drink too much. Many spend most of the day sitting. Those things have real consequences for testosterone, sperm quality, erections, fertility, and long-term health. The good news is that these are not fixed traits. Lose weight if you need to. Build muscle. Train consistently. Sleep properly. Drink less. Treat sleep apnoea. Get outside. Improve insulin sensitivity. Stop smoking. Review medications with a doctor if they may be affecting hormones or fertility.

These interventions are not as viral as an apocalypse narrative. They are not as profitable as convincing millions of men they need immediate hormone treatment. But they are closer to the truth.