Vetenskap

The cannabis users had higher testosterone — not lower

By Daniel Johansson · Published 2026-08-06

The cannabis users had higher testosterone — not lower

A Swiss study published in April 2026 measured 70 endogenous steroids in the blood of 47 young cannabis consumers and 47 controls. Testosterone was higher in the users, not lower. So were androstenedione and DHT. The adrenal hormones were unchanged. It is a clear finding — and it does not say what is being said about it on social media.

What was actually measured

The study comes from the University of Geneva, together with the Swiss Centre for Applied Human Toxicology in Basel and the clinical chemistry laboratory at the Hospital of Valais in Sion. The participants are Swiss men aged 18–23 at the time of sampling, recruited nationwide through mandatory military enrolment between 2005 and 2017.

The method is liquid chromatography with tandem mass spectrometry (LC-MS/MS). 171 steroids were targeted; 70 met the quality criteria and are reported — 17 androgens, 15 progestogens, 3 estrogens, 20 corticosteroids, 3 oxysterols, 9 glucuronides, 5 sulfates and 5 bile acids. Seven of the major steroids were absolutely quantified against ¹³C-labelled internal standards.

Group allocation does not rest on declared use alone, but on what was in the blood. The consumer group had declared use within the previous seven days and had measurable serum THC and THC-COOH; the control group had neither. That is the study's strength and simultaneously its most important limitation: detectable serum THC corresponds to use within roughly twelve hours, derived through a published pharmacokinetic model. Every consumer in the study is therefore a current user, not a former one.

Fourteen of the 47 are classified as chronic consumers (THC-COOH above 40 µg/L), 33 as occasional.

What went up — and what did not

Seven of the steroids were absolutely quantified. Those are what the study's Table 1 reports, in nmol/L:

The p-values come from two-tailed t-tests without adjustment for multiple comparisons. That is stated in the study's own table heading, and it is worth carrying: with 70 steroids measured, individual p-values just under 0.05 are weaker evidence than they look.

The testosterone difference works out at 23% on the published means. That percentage does not appear in the study — it is calculated from the table, and should be read as what it is: a difference between two group means in a cross-sectional study with 47 people per arm.

The remaining steroids were measured relatively rather than absolutely. There: dihydrotestosterone (DHT) was significantly higher in the THC-positive men (p = 0.029). The C11-oxy androgens — 11β-hydroxytestosterone, 11β-hydroxyandrostenedione, 11-ketotestosterone and 11-ketoandrostenedione — showed no difference. LH and FSH did not differ significantly between the groups either.

The strongest finding in the multivariate model was not testosterone at all, but two progesterone metabolites: 5β-dihydroprogesterone (p = 1.9 × 10⁻⁵) and 11β-hydroxyprogesterone (p = 5.7 × 10⁻⁵). The first showed a dose-dependent relationship between chronic and occasional users (p = 0.0027); the second behaved as a general exposure marker.

Why the pattern points at the testes

It is the combination of what rose and what did not that carries the interpretation.

The androgens produced in the testes were elevated. The androgens carrying the adrenal signature — the C11-oxy androgens — were not. Cortisol and cortisone were unchanged. The authors therefore conclude that the alterations are confined to the gonadal part of the HPG axis, supporting a possible direct effect of phytocannabinoids on testicular sex hormone synthesis, mediated by CB1 receptors in Leydig cells.

But the route there is not established. The study could find no relation between LH or FSH levels and the sex steroids, and neither pituitary hormone differed between the groups. The authors give two reasons why that does not settle the question: LH and FSH are secreted in pulses, which makes single-time-point measurements hard to compare, and GnRH — the next step up — does not circulate in the blood and cannot be measured at all. Their own formulation is that the precise mechanisms remain unclear.

Three other explanations the authors raise themselves

What makes the study worth reading is that it does not stop at its own finding. Three alternatives remain standing in the discussion, and none can be dismissed on the available evidence:

Reverse causality. Men with naturally higher testosterone may be more prone to cannabis use, through greater risk-taking. In that case it is not the cannabis that raised the testosterone, but the testosterone that led to the cannabis. The authors weigh it down themselves — in the literature they cite, the link between testosterone and risk-taking is described as significant but modest, and possibly overstated when social status is not accounted for. But they do not strike it.

The liver. CB1 and CB2 receptors are expressed in the liver, and an effect of phytocannabinoids on hepatic steroid metabolism cannot be ruled out in this study, the authors write.

Compensation. The raised testosterone may be a response rather than an effect: a homeostatic compensation for reduced androgen receptor sensitivity in the presence of phytocannabinoids. If that holds, a higher value in the vial does not mean more androgen signal in the body — it means the body has to turn the volume up to reach the same effect.

The 1974 study — and why it was not simply wrong

The widespread belief that cannabis lowers testosterone traces to a single paper: Kolodny and colleagues in the New England Journal of Medicine, April 1974. Twenty men who had smoked at least four days a week for at least six months were compared with twenty controls. Testosterone was markedly lower in the users, 416 against 742 ng per 100 mL.

The authors of the 2026 study point to that paper themselves as the only one reporting a significant decrease, noting that it rested on a small cohort and lacked control for key confounding factors.

But the story that one faulty paper governed fifty years of research does not hold all the way. Seven months later, in November 1974, Mendelson and colleagues published a measurement of testosterone levels before, during and after chronic cannabis smoking in the same journal. That the study exists, and arrived immediately, is established; what it concluded Svenskhampa has not been able to verify against full text. What is established is the 2026 authors' own formulation: Kolodny stands alone in reporting a significant decrease. Fifty years of research did not confirm it. It simply won the distribution.

There is also a difference in exposure that skews the comparison in both directions. Kolodny measured chronic heavy consumers over months. The 2026 study measures people with THC in the blood right now. These are not the same question, and they need not give the same answer.

Higher testosterone is not the same as better fertility

This is the point where interpretation most often derails, and it has already been demonstrated in a considerably larger study.

Gundersen and colleagues examined 1,215 healthy Danish men aged 18–28, recruited at military enrolment between 2008 and 2012, publishing in the American Journal of Epidemiology in 2015. The cannabis smokers had higher testosterone — the same direction as in the Swiss study. The Swiss group cites the figure themselves: 7% higher after adjustment for confounders.

At the same time, those smoking more than once a week had 28% lower sperm concentration (95% CI: −48, −1) and 29% lower total sperm count (95% CI: −46, −1) after adjustment. Combining cannabis with other recreational drugs more than once a week produced reductions of 52% and 55% respectively.

Two things deserve reading out properly. The confidence intervals run all the way up to −1%, almost to zero — the finding is statistically significant but far from precise. And the raised testosterone sat, according to the authors, within the same range as in cigarette smokers, which makes it hard to read as anything cannabis-specific.

But the direction is what matters: higher testosterone and worse semen quality, in the same cohort, in the same men. That is exactly what makes "cannabis raises testosterone" a poor summary of the evidence — the claim is right on the facts and wrong on the meaning.

What this is not

• It is not a causal finding. The study is cross-sectional. It shows the groups differ, not that cannabis caused the difference. The authors write themselves that the number of factors influencing steroid metabolism carries an inherent risk of false positives.
• It is not an effect of long-term use. Every THC-positive participant had used cannabis within roughly twelve hours. The study cannot classify them more finely, and the authors state it permits no extrapolation to long-term effects or to past users.
• It does not apply to everyone. The basis is 47 Swiss men aged 18 to 23. The authors state explicitly that the results should not be generalised to women, other age groups or more diverse populations.
• It says nothing about CBD or hemp products. Participants were split on serum THC. Nothing in the study concerns cannabidiol, hemp seed or industrial hemp foods, and no conclusions about such products follow from it.
• Alcohol, diet, sleep and stress could not be controlled. BMI, sampling time and tobacco were examined and did not differ significantly between groups. Remaining lifestyle factors could not be assessed, which the authors report.
• The HPG axis was not measured directly. GnRH does not circulate in the bloodstream and could not be evaluated. LH and FSH are secreted in pulses, which the authors note makes single-time-point measurements hard to compare. The proposed mechanism is exactly that — proposed.
• The p-values are unadjusted. The study's own table heading specifies two-tailed t-tests without adjustment for multiple comparisons. With 70 steroids measured, individual findings near 0.05 — such as 17α-hydroxyprogesterone at 0.03 — should be read cautiously. The two progesterone metabolites at the 10⁻⁵ level are untouched by the objection.
• The 1974 papers were not read in the original. The Kolodny figures are relayed via the 2026 article and via a published abstract. Mendelson's paper is established as to title, journal and date, but its content has not been verified against full text. The 2026 article and Gundersen's abstract were both read in full.

Sources

• Galmiche M, Meister I, Zufferey F, Rahban R, Senn A, Boccard J, Rossier MF, Nef S, Rudaz S. Cannabis consumption is associated with altered steroid metabolism in young men. Commun Med (Lond) 2026 Apr 16;6:224. doi:10.1038/s43856-026-01469-x · full text at PubMed Central
• Kolodny RC, Masters WH, Kolodner RM, Toro G. Depression of plasma testosterone levels after chronic intensive marihuana use. N Engl J Med 1974;290:872–874. PubMed 4816961
• Mendelson JH, Kuehnle J, Ellingboe J, Babor TF. Plasma testosterone levels before, during and after chronic marihuana smoking. N Engl J Med 1974 Nov 14;291(20):1051–1055. PubMed 4415097
• Gundersen TD, Jørgensen N, Andersson AM, Bang AK, Nordkap L, Skakkebæk NE, Priskorn L, Juul A, Jensen TK. Association Between Use of Marijuana and Male Reproductive Hormones and Semen Quality: A Study Among 1,215 Healthy Young Men. Am J Epidemiol 2015;182(6):473–481. PubMed 26283092

This article references published research for journalistic purposes. It is not medical advice, and no health claims about any product follow from it. See Ownership and transparency for Svenskhampa's relationship with Helsama.