Testosterone governs wanting more than it governs working.
Testosterone has a real and consistent effect on desire. Its effect on erections is smaller, and concentrated in men who are genuinely deficient.
Testosterone governs wanting more than it governs working.
That single distinction explains most of what confuses men about hormones and sex — why a man on testosterone can find his interest returns while his erections don’t, why sildenafil works when testosterone doesn’t, and why the largest trial ever conducted found improvement in one and not the other.
Four systems, not one
“Sexual function” bundles together four things that run on different machinery.
Desire — wanting sex. Substantially androgen-dependent. This is testosterone’s clearest domain.
Arousal and erection — a vascular and neurological event. Nitric oxide, smooth muscle relaxation, arterial inflow, venous trapping. Testosterone supports the tissue, but the mechanics are largely plumbing.
Orgasm — a spinal and cerebral reflex, poorly understood, weakly linked to testosterone.
Ejaculation — a separate reflex again, with its own neurology.
A man can have any combination: strong desire with unreliable erections, reliable erections with no interest, both intact with delayed orgasm. Treating all four as one dial is the error that produces disappointed patients.
What the trials actually found
Two large studies point in what looks like opposite directions, and reconciling them is the most useful thing on this page.
The Testosterone Trials, in men with unambiguous deficiency, found significant improvement across all three sexual domains: sexual activity with a treatment effect of 0.58 (95% CI 0.38–0.78), sexual desire 2.93 (95% CI 2.13–3.74), and erectile function 2.64 (95% CI 1.68–3.61) — all p<0.001.
TRAVERSE, in 5,246 older men at high cardiovascular risk, found sexual activity improved by 0.49 acts per day at six months (95% CI 0.19–0.79) and desire improved significantly — with no significant difference in erectile function (p=0.443; a 0.6-point between-group difference at twelve months, and −0.2 at twenty-four).
Desire improved in both. Erections improved in one and not the other.
The reconciliation is baseline severity. A 2024 meta-analysis by Xu and colleagues, pooling 12 randomized trials in 1,466 men, found testosterone improved erectile function scores by a weighted mean of 3.26 IIEF points (95% CI 1.65–4.88, p<0.0001). The severity gradient comes from a separate and larger meta-analysis — Corona and colleagues in European Urology, 14 placebo-controlled trials in 2,298 men — which found the improvement was 2.95 points (95% CI 1.86–4.03) in men below 8 nmol/L versus 1.47 points (95% CI 0.90–2.03) in men below 12 nmol/L, a significant difference between subgroups (Q=5.61, p=0.02). That paper includes Abraham Morgentaler among its authors, which is worth knowing given his advocacy in this field; the finding is nonetheless the best-powered subgroup analysis available. The benefit concentrates in genuine, marked deficiency and fades toward nothing in the borderline range.
TRAVERSE enrolled men with milder deficiency, an average age around 65, and a heavy burden of diabetes and vascular disease. Most of their erectile dysfunction was not hormonal, so correcting the hormone did not fix it.
What this means for you: the more clearly deficient you are, the more likely testosterone helps your erections. If your testosterone is 320 and you are 60 with hypertension and diabetes, the honest expectation is that your desire may improve and your erections probably won’t.
Adding testosterone to sildenafil
A reasonable idea: if a man with low testosterone doesn’t respond adequately to a PDE5 inhibitor, does adding testosterone rescue it?
The best trial says no. Spitzer and colleagues, publishing in Annals of Internal Medicine in 2012, took around 140 men aged 40 to 70 with erectile dysfunction and low testosterone, optimized them on sildenafil first, then randomized them to testosterone gel or placebo for fourteen weeks. The primary erectile function outcome showed a mean difference of 2.2 points (95% CI −0.8 to 5.1, p=0.150) — not significant. Desire, intercourse frequency and satisfaction were similarly unmoved. Sildenafil was doing the work.
There is older, smaller, positive literature — a 2004 randomized trial reported benefit from adding testosterone in hypogonadal sildenafil non-responders. It is smaller and of lower quality than Spitzer. The honest summary is that the evidence is mixed and the best trial is negative.
Evidence: 🟠 Limited, trending negative
Morning erections
Loss of morning erections is one of the three symptoms the European Male Ageing Study found independently associated with testosterone, alongside low desire and erectile dysfunction. That makes it worth mentioning to a clinician.
It is not a test. We could not find sensitivity or specificity figures, and morning erections vary with sleep quality, alcohol, age and medication. Their absence is a reason to investigate, not a diagnosis.
Erectile dysfunction is a cardiovascular warning
This is the most important thing on this page and it has nothing to do with testosterone.
The penile arteries are small. When endothelial disease begins, they narrow before the coronary arteries produce symptoms. Erectile dysfunction is often the first clinically visible sign of vascular disease.
A meta-analysis of 12 prospective cohort studies published in the Journal of the American College of Cardiology found erectile dysfunction associated with a relative risk of 1.48 (95% CI 1.25–1.74) for cardiovascular disease, 1.46 (1.31–1.63) for coronary heart disease, 1.35 (1.19–1.54) for stroke, and 1.19 (1.05–1.34) for all-cause mortality.
In 300 consecutive men with chest pain and angiographically confirmed coronary disease, erectile dysfunction was present in 49%, and in 67% of those who had both, it appeared before the cardiac symptoms — by a mean of 38.8 months, a little over three years (Montorsi and colleagues, European Urology, 2003). That is a referral-based angiographic series rather than a population cohort, so read the interval as indicative rather than precise. The association itself is solid, and it is strongest in younger men: in a population cohort followed for a decade, men aged 40 to 49 with erectile dysfunction had a coronary disease incidence roughly fifty times that of men the same age without it, while by age 70 the difference had almost vanished (Inman and colleagues, Mayo Clinic Proceedings, 2009).
A man who develops erectile dysfunction in his forties or fifties should have his cardiovascular risk assessed. Blood pressure, lipids, glucose, smoking, weight. Not only his testosterone.
This is the single most valuable thing a man can take from this chapter, and it is the thing a clinic selling testosterone has the least incentive to tell him.
Psychogenic, vascular, hormonal
Broadly: psychogenic causes predominate in younger men — performance anxiety, situational or relationship-related, often with preserved morning and masturbatory erections. Vascular and other organic causes predominate with age and comorbidity, typically with a gradual onset and loss of morning erections. Hormonal causes are the smallest slice and usually present with reduced desire alongside.
We could not verify percentage breakdowns and are not publishing any. The pattern-recognition above is well established clinically; the numbers behind it were not accessible.
The practical distinction: sudden onset with preserved morning erections and situational variability points away from a vascular cause. Gradual onset, absent morning erections, cardiovascular risk factors points toward one.
What remains uncertain
Whether 8 nmol/L is a threshold or just a convenient cut point. The Corona meta-analysis shows a significantly larger effect below 8 nmol/L than below 12, but subgroup boundaries chosen for analysis are not the same thing as a biological switch, and nobody has demonstrated one.
Whether testosterone improves PDE5 inhibitor response in any subgroup. The best trial is negative; older smaller trials disagree.
Whether TRAVERSE’s null erectile finding reflects population or a genuine ceiling on what testosterone can do for erections.
How much erectile dysfunction in middle-aged men is vascular versus psychogenic versus hormonal. Not verifiable to a publishable standard.
Whether testosterone affects orgasm, ejaculate volume or “stamina.” A negative finding for ejaculatory dysfunction exists in the literature; we could not confirm the details. The pattern across trials — desire and activity respond, mechanics largely don’t — suggests little effect, but that is inference.
A note on the other half of the bed
Testosterone’s clearest effect in men is on desire rather than on erections. The same is true in women: the one indication with good randomised evidence behind it is distressing low sexual desire after menopause, not energy or ageing. Two separate evidence bases, in two sexes, converge on the same conclusion about what this hormone actually does.
Couples often arrive at this subject together, and the questions are not symmetrical. What testosterone is for in women, what the evidence supports, what it does not, and what a man using a topical gel needs to know about transferring it to a partner are covered in testosterone in women.
Questions patients ask
Low testosterone causes erectile dysfunction.
It contributes in markedly deficient men. Most erectile dysfunction is vascular.
What the evidence showsErectile dysfunction was one of three symptoms independently associated with testosterone in the European Male Ageing Study. But TRAVERSE found no erectile improvement, and meta-analysis shows the benefit concentrates in men below about 8 nmol/L.
What remains uncertainThe exact threshold at which erectile function becomes androgen-dependent.
Bottom lineWorth measuring testosterone. Don't expect it to be the answer.
Moderate
TRT will cure my ED.
Usually not — and the largest trial is unambiguous on this.
What the evidence showsTRAVERSE found no significant difference in erectile function (p=0.443) despite improvements in desire and sexual activity. The Testosterone Trials, in more clearly deficient men, did find improvement (2.64, 95% CI 1.68–3.61).
What remains uncertainWhich individual men fall into the responsive group.
Bottom lineIf your erections are the problem, testosterone alone is a poor bet.
Strong
Testosterone improves erections.
In severely deficient men, modestly. In borderline men, barely.
What the evidence showsXu and colleagues (2024), pooling 12 trials in 1,466 men, found a weighted mean improvement of 3.26 IIEF points (95% CI 1.65–4.88). Corona and colleagues (*Eur Urol* 2017), pooling 14 trials in 2,298 men, found 2.95 points below 8 nmol/L versus 1.47 below 12 (Q=5.61, p=0.02).
What remains uncertainWhether a 3-point IIEF change is what a man would call meaningful.
Bottom lineReal but modest, and concentrated in genuine deficiency.
Moderate
Why does Viagra work when testosterone doesn't?
Because they fix different problems.
What the evidence showsPDE5 inhibitors act on the vascular mechanism of erection directly. Testosterone acts primarily on desire. Most erectile dysfunction in middle-aged and older men is vascular.
What remains uncertainNothing about the mechanism.
Bottom lineDifferent machinery, different drug.
Strong
Adding testosterone improves how well Viagra works.
The best trial says no.
What the evidence showsSpitzer et al., *Annals of Internal Medicine* 2012, randomized around 140 men with low testosterone and erectile dysfunction, already optimized on sildenafil, to testosterone or placebo. Mean difference 2.2 points (95% CI −0.8 to 5.1, p=0.150) — not significant, with desire and satisfaction similarly unchanged. Older, smaller positive trials exist.
What remains uncertainWhether a more deficient subgroup would respond.
Bottom lineMixed evidence, best trial negative. Not a reason to add testosterone.
Limited
Low testosterone causes low libido.
This is testosterone's strongest and most consistent effect.
What the evidence showsLow sexual desire was one of the three symptoms independently associated with testosterone in EMAS. The Testosterone Trials found a desire treatment effect of 2.93 (95% CI 2.13–3.74, p<0.001), and TRAVERSE found significant improvement in desire.
What remains uncertainHow much desire is hormonal versus relational, psychological or situational in any individual.
Bottom lineThe one sexual outcome testosterone reliably moves.
Strong
You can't have high testosterone and low libido.
You certainly can, and it's common.
What the evidence showsDesire is affected by relationship factors, depression, medications including SSRIs, stress, sleep, chronic illness and pornography-related conditioning. Testosterone is one input among many.
What remains uncertainThe relative weight of these factors at population level.
Bottom lineNormal testosterone with low desire means the cause is elsewhere. More testosterone won't help.
Moderate
Masturbation lowers testosterone.
No. This is the most persistent myth in men's health and it traces to one small retracted study.
What the evidence showsThe claim comes from a 2003 study of 28 men reporting a single transient peak at day seven of abstinence, reaching about 145.7% of baseline, which did **not** persist with continued abstinence. The paper was subsequently **retracted** — for duplicate publication, not fraud. It was small, unblinded and unreplicated. Fact-checkers have rated the popular "45% testosterone increase from seven days" claim false.
What remains uncertainWhether the transient day-seven blip is even real, given the study's fate.
Bottom lineNo credible evidence that masturbation lowers testosterone or that abstinence raises it durably.
Strong (that the claim is unsupported)
Ejaculation lowers testosterone.
Same answer, same non-existent evidence base.
What the evidence showsNo credible evidence links ejaculation frequency to lasting testosterone changes in normal men. The literature cited for this consists of the retracted 2003 study.
What remains uncertainNothing that would change the practical answer.
Bottom lineEjaculation frequency is not a testosterone variable.
Strong (that the claim is unsupported)
Abstinence increases testosterone.
Not durably, and the evidence for even a transient effect is a retracted paper.
What the evidence showsThe single transient day-seven peak reported in the 2003 study did not persist, and that paper was retracted. Nothing else supports a sustained rise.
What remains uncertainWhether men who report feeling better during abstinence are experiencing something real that isn't hormonal.
Bottom lineIf abstinence helps you in other ways, fine. It isn't your testosterone.
Strong (that the claim is unsupported)
Testosterone increases penis size.
Not in adults.
What the evidence showsAndrogens drive genital development in fetal life and puberty. There is no randomized or solid population evidence that circulating testosterone changes penis size in men who have already developed. The pediatric androgen-stimulation literature does not generalize to adults.
What remains uncertainNothing meaningful.
Bottom lineIt does not. Any clinic implying otherwise is selling something.
Strong
Testosterone improves orgasm.
No good evidence, and the pattern across trials suggests little effect.
What the evidence showsTrials consistently show testosterone moving desire and sexual activity while leaving the mechanics — erection specifically — unchanged. Orgasm-specific outcomes are poorly studied.
What remains uncertainOrgasm quality is barely measured in these trials.
Bottom lineNot something to expect.
Limited
Testosterone improves sexual stamina.
No evidence supports it.
What the evidence shows"Stamina" isn't a measured endpoint. Ejaculatory latency is governed by different neurology, and a negative finding for testosterone in ejaculatory dysfunction exists in the literature, though we could not confirm its details.
What remains uncertainAlmost everything, because it isn't studied.
Bottom lineNot a documented effect.
Limited
Testosterone increases penile sensitivity.
No reliable evidence.
What the evidence showsNot a measured outcome in the major trials. Nothing to cite.
What remains uncertainAll of it.
Bottom lineUnsupported.
Unsupported
Morning erections tell you your testosterone is fine.
Their loss is a clue. Their presence is not a clearance.
What the evidence showsReduced morning erections was one of three symptoms independently associated with testosterone in EMAS. We could not find sensitivity or specificity figures, and morning erections vary with sleep, alcohol, age and medication.
What remains uncertainHow well they perform as a marker.
Bottom lineWorth mentioning to your clinician. Not a substitute for a blood test.
Limited
Erectile dysfunction can be the first sign of heart disease.
Yes — and this is the most important sentence on this page.
What the evidence showsA meta-analysis of 12 prospective cohorts in the *Journal of the American College of Cardiology* found erectile dysfunction associated with a relative risk of 1.48 (95% CI 1.25–1.74) for cardiovascular disease, 1.46 for coronary heart disease, 1.35 for stroke, and 1.19 for all-cause mortality. The penile arteries are smaller than the coronaries and show disease earlier.
What remains uncertainThe commonly quoted three-year lead time — we could not verify it from the primary source.
Bottom lineNew erectile dysfunction in a man in his forties or fifties warrants a cardiovascular workup, not just a hormone panel.
Strong
Where to go next
Cardiovascular risk in detail: Testosterone, the Prostate and the Heart What testosterone actually does: What Testosterone Actually Does If something else explains your symptoms: When Testosterone Isn’t the Answer
Ready for testosterone care built on a diagnosis?
The first step is a full endocrine evaluation, not a prescription. We see patients across San Diego County and welcome referrals from other physicians.