Testosterone Test: What Time to Have Blood Drawn and Why You Must Repeat the Result

Testosterone secretion follows a circadian rhythm. In young men the concentration rises at night during sleep, peaks in the early morning hours and then gradually declines through the day, reaching its lowest in the evening. The rhythm is driven by pulsatile LH secretion from the pituitary and is linked to sleep. That is why the same man's value at 8 a.m. and at 4 p.m. can be clearly different.

In a study in which blood was drawn every 30 minutes for a whole day in 10 young and 8 middle-aged men, both groups showed a significant circadian rhythm for all measured variables, and the fall in total testosterone from peak to nadir was at least 43% in every participant. The peak (acrophase) of total, free and bioavailable testosterone fell between 7:00 and 7:30 a.m. For SHBG it was the opposite: the peak fell in the afternoon (about 3:00–3:30 p.m.), which additionally affects free testosterone results during the day.

What does this mean in practice? Testosterone is 'sensitive to the clock' more than most other reproductive-axis hormones. In the Brambilla et al. study, for dihydrotestosterone, SHBG, LH, FSH and estradiol the diurnal variation during laboratory hours was much smaller, so those hormones can be measured at any time of day, while testosterone cannot.

For practice two numbers matter most: 20–25% (the fall between eight and four o'clock in men aged 30–40) and 17 men in whom a single afternoon result fell below 300 ng/dl although all morning results were normal. For a man with a borderline result, e.g., 350 ng/dl in the morning, moving the test to the afternoon can 'produce' a false diagnosis of deficiency.

It is often repeated that in older men the circadian rhythm 'disappears', so the time of collection does not matter. The data are less clear. In the Boston study the difference between morning and afternoon declined with age (about 10% at age 70) but did not vanish, and in nine men aged 66–80 a single low afternoon result occurred despite normal morning results. In a 24-hour study in healthy, fit middle-aged men, the circadian rhythm of testosterone was preserved and did not differ significantly from young men in parameters such as the mean or phase.

A cautious conclusion: in older healthy men the rhythm may be somewhat weaker but is still large enough that an afternoon measurement could lead to a false diagnosis. The 'morning' recommendation therefore applies at every age. An acceptable compromise for people who cannot come early is collection as early as possible, but before noon — and sticking to the same hour on repeats.

This study teaches two things. First, even with perfectly observed time and collection conditions, the same man's result can differ by a dozen or so percent from day to day — hence the need to confirm a low value. Second, a change in result from 400 to 480 ng/dl or from 520 to 450 ng/dl on a check after a few weeks usually means neither improvement nor worsening; it falls within natural 'noise'.

There is also variation between laboratories. Harmonized reference ranges for non-obese men aged 19–39, developed in four large cohorts and calibrated to the CDC reference method, were 264–916 ng/dl (2.5th–97.5th percentile), and the authors showed that a substantial part of the differences between cohorts stems from differences between assays. That is why results from different laboratories are hard to compare, and a follow-up is best done in the same place.

The 2018 Endocrine Society guideline recommends diagnosing hypogonadism only in men with symptoms and unequivocally, consistently low testosterone. As the initial test it recommends measuring total testosterone in the morning, fasting, with a reliable and accurate method, and then confirming the diagnosis by repeating the fasting morning measurement. In men whose total testosterone is near the lower limit of normal or who have a condition that alters SHBG, it recommends measuring free testosterone by equilibrium dialysis or calculating it with an accurate formula.

Note the wording: the result is to be 'unequivocally and consistently' low. It is not a formality — it is real protection against over-diagnosis. Testosterone treatment is usually long-term, suppresses own production and fertility, can raise hematocrit and requires monitoring. Starting it on the basis of one imperfectly collected sample is an error that is easy to avoid.

The practical conclusion is simple: breakfast before the draw, and especially a sweet drink or carbohydrate-rich meal, can lower testosterone enough to spoil interpretation. The study used an oral load of 75 g of glucose, a stronger stimulus than a typical meal, so the fall after an ordinary breakfast may be smaller — but there is no reason to take the risk. Collection fasting (8–12 hours without food, with water) removes this factor. If a low result was obtained after a meal, it should be repeated fasting.

Incidentally, the same mechanism helps explain why men with insulin resistance and obesity often have lowered testosterone: part of this relationship is reversible with lifestyle change, as we write in the articles on insulin resistance and testosterone.

Looking for the 'best' result is similarly mistaken: doing the test several times and picking the highest to compare with a target or with people on forums. From a biological standpoint results are a distribution, not a single truth; what to look at is their reproducibility and whether symptoms coexist.

A good testosterone result starts before the laboratory: with the hour, with whether you drank coffee with sugar, and with whether you will do it a second time. The whole ritual is a few minutes and a few dozen zlotys more, and it saves you from two errors: unnecessary treatment after a result that turned out to be an artifact, and downplaying a problem when symptoms are real and a single result happens to be good.