Why Testosterone Studies in Women Keep Finding Nothing

If you've raised testosterone with a doctor and been told the evidence doesn't support it, you were told something true about the literature and something misleading about the hormone.

The trials that found nothing share a set of design problems, and once you can see them, the null results stop reading as an answer and start reading as a badly built question. They measured the hormone with a test that can't reliably read female concentrations. Most ran twelve weeks. Almost all of them asked about sexual function and nothing else.

None of that proves testosterone works for mood, drive, or anhedonia. I want to be careful about that, because the temptation to overclaim here is real. What it does mean is that the studies being quoted at you never actually tested the question you're asking.

Why the research missed it

The clinical research on testosterone in women was, in many cases, methodologically broken in ways that had nothing to do with whether the hormone works.

The problem is measurement quality. The studies used the wrong thermometer, measured the wrong thing, and stopped before the effects could establish themselves. What they found, or didn't find, tells us more about study design than it does about testosterone.

The measurement problem. The standard lab assay used to measure testosterone, called an immunoassay, performs far worse than its reputation suggests at female-range concentrations. The CDC ran an accuracy study that shows this plainly: 142 clinical laboratories were sent real human serum and asked to run it exactly the way they run a patient's sample. One specimen came from a woman whose true testosterone was 15.5 ng/dL by the reference method. The labs reported values ranging from 7.1 to 39.8 ng/dL. Grouped by the analyser they used, the averages spread from 12.7 to 26.8 ng/dL, and 44% of the results came back above 20 ng/dL. That's the same woman's blood. Several of the most widely used systems, the Beckman UniCel DxI 800 among them, missed every accuracy target the study set. That is not a measurement with error bars. That is essentially no reliable measurement at all. The Endocrine Society has recommended LC-MS/MS (liquid chromatography-tandem mass spectrometry) as the gold standard for measuring testosterone in women since 2007, eighteen years of official guidance that most commercial labs still don't act on by default. Both LabCorp (test code 070001) and Quest Diagnostics (test code 15983) now offer LC-MS/MS specifically for women; you may need to request it explicitly. Researchers were, in many cases, literally unable to determine whether two women had meaningfully different testosterone levels. A study designed to detect dose-response relationships in a population cannot find those relationships if it can't accurately measure the dose. You cannot conduct a valid thermometer study with a broken thermometer, and 7.1 to 39.8 ng/dL out of one woman's blood makes that metaphor literal.

Total vs. free testosterone. Studies almost uniformly measured total testosterone, the amount circulating in the blood. But total testosterone is not what your cells use. SHBG, sex hormone binding globulin, captures testosterone with high affinity and holds it unavailable. The biologically active form is free testosterone. A woman with high SHBG and "normal" total testosterone may have near-zero free testosterone and will feel every symptom of deficiency. Most studies never measured SHBG at all. The studies weren't measuring the thing that matters.

Studies ran too short. The majority of clinical trials on testosterone in women ran twelve to twenty-six weeks. Twelve weeks. Brain function, motivational circuitry, and dopaminergic tone are slow systems. The neuroscientists who study how testosterone modulates dopamine receptor density and tyrosine hydroxylase activity will tell you that meaningful neurological adaptation takes four to six months or longer to establish. A twelve-week study looking for changes in motivation and brain function is the equivalent of planting a tree in October and checking for fruit in December. The absence of fruit is not evidence that the tree doesn't bear fruit.

Worth knowing how short that really is by the field's own standard: the global consensus statement tells clinicians not to call testosterone a failure until six months have passed, and it only admitted trials of at least twelve weeks into the evidence base behind its own recommendations. Twelve weeks isn't a short trial. It's the floor.

The endpoints were wrong. In study after study, the primary outcome measure was sexual function. Sometimes specifically: number of satisfying sexual events per month. This is a legitimate outcome to measure. But it is a narrow slice of what testosterone does in the female body. Anhedonia, the loss of interest and pleasure in things you used to care about, was never a primary endpoint in any randomized trial. Motivation, creative drive, gym performance, willingness to initiate things, the quality of ambition, none of these were ever primary endpoints.

The 56% number. A 2024 to 2025 UK observational study of 510 women, the largest cohort ever assembled to look specifically at non-sexual testosterone outcomes, found something remarkable. When women reported symptom improvement on testosterone therapy, the symptom with the highest response rate wasn't libido. It was anhedonia. Fifty-six percent of women in the cohort reported improvement in anhedonia, loss of interest in things they used to care about. That is the highest response rate of any symptom measured. No randomized controlled trial has ever used anhedonia as a primary endpoint. The thing women experience most powerfully was never the thing researchers were measuring.

The placebo problem. In testosterone studies, anywhere from 31 to 49 percent of women in placebo arms report symptomatic improvement. This is not unusual for trials involving subjective symptom measures, it reflects the complexity of expectation, therapeutic attention, and the natural variability of symptoms over time. But it means you need very large studies, very precise measurement tools, and appropriate study duration to detect the signal above the noise. Most testosterone studies were not designed that way. They were underpowered, used imprecise assays, ran too short, and measured the wrong endpoints.

Put it together: the research didn't fail to find robust effects because robust effects don't exist. It failed to find robust effects because it used a broken thermometer, measured the wrong thing, for too short a time, in too small a population, with the wrong outcomes. That's evidence of poor study design.

The 56% anhedonia response rate in 510 women, from a single observational study, is more clinically meaningful than most of the randomized trial literature, because it's the first time someone asked the right question with a large enough sample to see the answer.

To be clear about the evidence: the sexual-function benefit is proven in randomized trials. The mood, drive, and anhedonia benefits are supported so far by mechanism, one large observational cohort, and clinical experience, not yet by placebo-controlled trials designed to measure them.

And your own level may be lower than your result suggests

The same measurement problem shows up in your own lab work, not just in the research. Oral estrogen and the combined pill drive SHBG up, and SHBG binds testosterone tightly enough that your total can read perfectly normal while very little of it is available to your tissues.

The size of that effect is not subtle. In a 2022 randomized trial, a pill containing ethinylestradiol raised SHBG by 385% and cut the free androgen index by 72%, while total testosterone didn't move at all. The test your doctor ordered stayed reassuringly normal while the testosterone your cells could actually use dropped by nearly three quarters.

That has its own page, including what happens after you stop the pill, which turns out to be faster than I originally wrote: birth control and testosterone, why your total reads normal and your free does not.

What to ask your doctor: Request total testosterone, free testosterone (specify equilibrium dialysis, the most accurate method), and SHBG together, as a panel. A morning blood draw gives you peak levels. This isn't exotic testing; it's just asking for the right thing.

This is the gap the whole site sits in. Not a proven therapy being withheld from women, which would be a simpler and angrier story, but a question the research has never asked properly, in a population it was not built to measure, using a test that cannot reliably read the numbers involved.

That distinction matters, because it changes what you're asking your doctor for. You're not asking them to ignore the evidence. You're asking them to know what the evidence actually measured.

This is one of four pieces on the same subject. The others: what testosterone does, and what happened when I took it · the side effects, honestly · why there's no FDA-approved product for women

Sources

Why the labs fail women - Rosner W et al., J Clin Endocrinol Metab 2007: the Endocrine Society position statement recommending LC-MS/MS as the gold standard for measuring testosterone. - Cao Z et al., Clinica Chimica Acta 2017: the CDC accuracy-based proficiency study. 142 laboratories, unaltered single-donor human serum run as routine patient samples. The female specimen with a reference value of 15.5 ng/dL was reported back as anything from 7.1 to 39.8 ng/dL. - Haverinen A et al., Journal of Clinical Endocrinology & Metabolism 2022: randomized trial. A pill containing ethinylestradiol raised SHBG by 385% and cut the free androgen index by 72%, while total testosterone was unchanged. - Hugon-Rodin J et al., Blood 2024: prospective cohort. After stopping combined hormonal contraceptives, SHBG fell rapidly, about 76% of the decline within two weeks, and by twelve weeks was no different from women who had never used them.

What the trials did and didn't ask - Davis SR et al., Journal of Sexual Medicine 2019, the Global Consensus Position Statement on Testosterone Therapy for Women. Its recommendations rest only on blinded randomized trials of at least twelve weeks, and it advises ceasing treatment if no benefit is experienced by six months. - Islam RM et al., Lancet Diabetes and Endocrinology 2019, the definitive meta-analysis of randomized trials: testosterone reliably improves sexual function in women. It also found the evidence on mood, cognition, and musculoskeletal outcomes was still insufficient, which is exactly the gap this article is about.

Beyond libido - Glynne S, Newson L et al., Archives of Women's Mental Health 2024, an observational cohort of 510 women on transdermal testosterone: the most-improved symptom was loss of interest in most things (anhedonia) at 56%. Observational and open-label, not a placebo-controlled trial, but the largest look yet at non-sexual outcomes.