That is what the Endocrine Society recommends [1]. The catch is that the two guidelines American men are usually tested under disagree on fasting. The American Urological Association calls the evidence "particularly weak" and tells urologists not to insist on it [2]. We think the endocrinologists have the better argument. A test that decides whether you start a lifelong prescription is worth one skipped breakfast.
The practical answer: timing, fasting and the repeat
- Fast from the night before. The Endocrine Society recommends a fasting morning total testosterone, confirmed by a second one [1]. No guideline names an exact number of hours. The meal studies below used overnight fasts of 10 to 12 hours [4], so finishing dinner by 8 or 9 p.m. matches the research.
- Water is fine; skip the coffee. No study we found tested either. The authors of the main meal study advise "a complete overnight fast" and note that patients often ask about coffee or juice [4]. Water carries no glucose. Coffee is untested, which is not the same as safe. Drink it after the draw.
- Get drawn early. Both guidelines want early-morning samples [1][2]. Before 10 a.m. is a sensible target; younger men should aim closer to 7 a.m., for reasons covered below.
- Test twice, on different days. The AUA's strongest diagnostic statement — Strong Recommendation, Grade A evidence — is that low testosterone is diagnosed "only after two total testosterone measurements are taken on separate occasions" in the early morning [2]. The Endocrine Society cites data that roughly 30% of men with an initial reading in the hypogonadal range are normal on repeat [1].
- Same lab, same assay. The AUA asks clinicians to use the same laboratory and method for serial measurements [2]. Switching between a mail-in kit and a hospital lab between tests adds noise you cannot separate from a real change.
The AUA uses 300 ng/dL as its cut-off for low total testosterone [2]. The Endocrine Society points to a harmonized reference range of 264 to 916 ng/dL in healthy, nonobese men aged 19 to 39, measured with assays calibrated to the CDC reference method [1][11]. Neither number diagnoses anything on its own. Both guidelines also require symptoms [1][2]. If you are trying to work out whether your symptoms are low T or simply age, our guide to low testosterone vs normal aging covers that side.
Why food and glucose suppress testosterone
The fasting advice is not folklore. Three studies measured it directly, and they point the same way.
The glucose study. At Massachusetts General Hospital, 74 men aged 19 to 74 drank the standard 75 g glucose solution used for diabetes testing. Mean testosterone fell 25%, and it was still down at two hours: 13.7 nmol/L compared to 16.5 at baseline [3]. Of the 66 men who started in the normal range, 10 (15%) dropped into the hypogonadal range at least once during the test. The effect was the same whether or not the men were diabetic or overweight [3].
The mixed-meal study. A Boston group measured testosterone with liquid chromatography–tandem mass spectrometry, the most accurate method available, in 60 men with normal levels and no diabetes [4]. After a glucose drink, total testosterone fell by a mean of 100 ng/dL at its lowest point. After a liquid mixed meal, it fell by 123 ng/dL. The decline started 20 minutes in and bottomed out at about 60 minutes. About 11% of the glucose group and 56% of the meal group dipped below 300 ng/dL at some point [4]. That 56% needs context: the meal arm had only nine men. It is a striking number from a very small group, not a population estimate.
The Swedish breakfast study. After a standard 550-calorie meal, testosterone fell by 30% between 60 and 120 minutes compared to fasting samples from the same men [5]. LH and SHBG did not move.
That last detail matters for the mechanism. If eating lowered testosterone by changing the brain's signal to the testes (LH) or by changing the binding protein (SHBG), those would shift too. They did not. The MGH authors read this as a direct effect on the testes plus a central component, since LH failed to rise to compensate [3]. The honest answer is that nobody has pinned the mechanism down. The Swedish authors said as much in 2012 [5], and it is still the gap in this literature.
The AUA's objection is fair as far as it goes: these are small, short studies of drinks and liquid meals, not the bacon-and-eggs breakfast a man eats before a 9 a.m. appointment [2]. But none of them found eating harmless, and fasting costs one late breakfast. When the downside of a false low is a misdiagnosis and possibly years of therapy, fasting is clearly the better choice.
Why the clock matters as much as the meal
Time of day is the best-established factor. In a Boston study of 66 men aged 30 to 80, each tested three mornings and three afternoons, testosterone in men aged 30 to 40 was 20–25% lower at 4 p.m. than at 8 a.m. [6]. The gap narrowed with age to about 10% at 70. Among 17 men who had at least one afternoon reading under 300 ng/dL, nine were normal at every morning visit [6]. Those are men who would have been labelled low on an afternoon draw and were not.
There is one useful softening. A review of 2,569 men tested for erectile dysfunction found that only the youngest men showed a meaningful difference between 7–9 a.m. and 9 a.m.–2 p.m. draws [7]. The authors concluded that men under 45 should be drawn as close to 7 a.m. as possible, and men over 45 can be drawn any time before 2 p.m. without misleading results. Older men's rhythm flattens. If you are 52 and the only slot is 11 a.m., take it. If you are 34, book the first appointment of the day.
Then there is plain day-to-day noise. A study that drew men 1 to 3 days apart, then again at three and six months, found that two measurements of these hormones differed by more than 18–28% about half the time, depending on the hormone [8]. The authors concluded that one sample is not enough to characterise a man's levels, and that more than three is probably not worth it [8]. That is the case for the repeat test in one sentence.
What else skews the number
Fasting and timing do most of the work. These are the other things worth controlling, roughly in order of how much they matter.
- Acute illness. The Endocrine Society says not to test men who have, or are recovering from, an acute illness [1]. The AUA cites a small study of young men with respiratory infections in which mean total testosterone dropped 10%, and by up to 30% in some groups [2]. If you have had the flu in the past couple of weeks, reschedule.
- Short sleep. Ten healthy young men limited to five hours a night for a week saw daytime testosterone drop 10–15% [9]. That was a small lab study, but the direction is consistent. A bad week of sleep before the test is a confounder, not a diagnosis. Our piece on TRT and sleep covers the relationship the other way round.
- Medications. Opioids and glucocorticoids lower testosterone, and the Endocrine Society flags both [1]. Tell whoever orders the test what you take.
- A workout right before the draw. Hard resistance training raises testosterone for 15–30 minutes afterwards [10]. That would push a reading up, not down, and the effect is brief. Evidence on whether a morning session meaningfully distorts a later draw is thin. The simple fix is to lift after the appointment, not before.
- The assay. Total testosterone measured by mass spectrometry is more accurate than immunoassay, especially at low levels, and the AUA prefers it where available [2]. The Endocrine Society wants a CDC-certified assay or one checked by an accuracy-based quality programme [1]. Much of the variation between large studies came from assay differences, not from the men [11].
- SHBG and free testosterone. Most testosterone in the blood is bound to SHBG. Obesity and type 2 diabetes lower SHBG, which can make total testosterone look low when the free fraction is fine. Older age and HIV raise it, which can do the opposite [1]. When total testosterone is near the lower limit, or SHBG is likely abnormal, the Endocrine Society recommends free testosterone by equilibrium dialysis or a validated formula. It says direct analog immunoassays for free testosterone are inaccurate and should not be used [1]. The AUA goes further: free testosterone should not be the primary diagnostic test, and total and free are not interchangeable [2].
If your numbers came back normal and you still feel off, that is a different problem, and our piece on why labs can look normal while you feel terrible is the place to start.
If you are already on TRT
On treatment the question changes from "is this low?" to "is the dose right?", and timing depends on the formulation [2]:
- Cypionate or enanthate injections: wait three to four cycles for levels to settle, then draw mid-cycle. The AUA admits there are no data on the best point in the cycle; mid-cycle is mostly convenience. A trough just before the next shot shows whether you stay therapeutic. A peak 18–36 hours after shows what drives side effects like a rising hematocrit.
- Long-acting undecanoate injections: test mid-cycle once steady state is reached, halfway between the first two 10-week injections.
- Pellets: test two to four weeks after the first implant to adjust the pellet count, then again at 10–12 weeks.
After that, every 6 to 12 months [2]. Our TRT best-practices guide covers the rest of the monitoring schedule.
Where to get tested
Online TRT services differ mostly in how the blood is drawn. PeterMD sends patients to LabCorp or Quest draw sites, which makes a fasting 7 a.m. venous draw easy to arrange. Maximus offers at-home lab testing. MangoRx and DudeMeds fold labs into the monthly price. Roman requires baseline labs and follow-up monitoring. Whatever the service, the same rules apply: fasted, early, twice. You can compare TRT providers side by side, and our hormone optimization hub covers in-person options.
Frequently asked questions
How many hours should I fast before a testosterone test?
No guideline gives an exact number. The Endocrine Society asks for a fasting morning sample [1], and the studies behind that advice used overnight fasts of 10 to 12 hours [4]. Finishing dinner the evening before and eating nothing until after a morning draw meets that standard.
Can I drink coffee before a testosterone blood test?
It has not been studied. The researchers behind the main meal study advise a complete overnight fast and specifically mention patients asking about coffee and juice [4]. Water is fine. Have the coffee after the draw.
Does eating really lower testosterone that much?
In controlled studies, yes, briefly. A glucose drink lowered testosterone by 25% on average [3], and a 550-calorie meal by 30% at 60 to 120 minutes [5]. In one study, testosterone started falling 20 minutes after eating [4]. It is temporary, but it is large enough to drag a borderline-normal man below 300 ng/dL.
What time of day is best for a testosterone test?
Early morning. In men aged 30 to 40, levels were 20–25% lower at 4 p.m. than at 8 a.m. [6]. Men under 45 should be drawn as close to 7 a.m. as possible; men over 45 can be drawn any time before 2 p.m. with little difference [7].
Why do I need two testosterone tests?
Because one is not reliable. The AUA requires two early-morning measurements on separate days to diagnose low testosterone [2]. About 30% of men with one low result are normal on repeat [1], and normal day-to-day variation is large [8].
Should I test free testosterone or total testosterone?
Total first. The Endocrine Society adds free testosterone when total is borderline or SHBG is likely abnormal, as in obesity, diabetes or older age, measured by equilibrium dialysis or a validated formula [1]. The AUA does not recommend free testosterone as the primary test [2].
Sources
- [1] Bhasin et al., Testosterone Therapy in Men With Hypogonadism: An Endocrine Society Clinical Practice Guideline, Journal of Clinical Endocrinology & Metabolism, 2018
- [2] Mulhall et al., Evaluation and Management of Testosterone Deficiency: AUA Guideline, American Urological Association, 2018 (validity confirmed 2024)
- [3] Caronia et al., Abrupt decrease in serum testosterone levels after an oral glucose load in men: implications for screening for hypogonadism, Clinical Endocrinology, 2013
- [4] Gagliano-Jucá et al., Oral glucose load and mixed meal feeding lowers testosterone levels in healthy eugonadal men, Endocrine, 2019
- [5] Lehtihet et al., S-testosterone decrease after a mixed meal in healthy men independent of SHBG and gonadotrophin levels, Andrologia, 2012
- [6] Brambilla et al., The effect of diurnal variation on clinical measurement of serum testosterone and other sex hormone levels in men, Journal of Clinical Endocrinology & Metabolism, 2009
- [7] Welliver et al., Validity of midday total testosterone levels in older men with erectile dysfunction, Journal of Urology, 2014
- [8] Brambilla et al., Intraindividual variation in levels of serum testosterone and other reproductive and adrenal hormones in men, Clinical Endocrinology, 2007
- [9] Leproult and Van Cauter, Effect of 1 week of sleep restriction on testosterone levels in young healthy men, JAMA, 2011
- [10] Kraemer and Ratamess, Hormonal responses and adaptations to resistance exercise and training, Sports Medicine, 2005
- [11] Travison et al., Harmonized reference ranges for circulating testosterone levels in men of four cohort studies in the United States and Europe, Journal of Clinical Endocrinology & Metabolism, 2017
Alpha Health Finder Editorial Team
Editorial Team
The Alpha Health Finder editorial team researches and writes evidence-based men's health content. Articles are grounded in cited primary sources and reviewed against a fixed editorial standard before publication. We are a research and directory team — not your prescribing clinicians.
Looking for an online provider?
Find a qualified online provider — compare options vetted and reviewed by Alpha Health Finder.
One email a week. The research, not the hype.
What we published this week, what we changed our minds about, and what it means if you are on a protocol.
No spam, no selling your address. Unsubscribe in one click.




