Normal Testosterone Levels by Age: What the Chart Hides

Almost every man who searches this phrase is holding a piece of paper. On it there is one figure and, beside it in smaller type, two more that mark out a range. If the figure sits inside the range there is a quiet sense of having been let off, and if it sits outside there is a jolt. Both reactions treat that printed range as though it meant something fixed.

It does not, and the reason is not a scandal. It is arithmetic. A reference range is a description of a group of people who were measured, using a particular machine, and both halves of that sentence move. This article is about how much they move, and what to do about it when the number in front of you is your own.

We earn nothing from any of this. There are no affiliate links on this site, nothing here is sponsored, and we do not sell tests or treatments, which is why this page can tell you that the reassuring range beside your result may have been drawn around a population that has nothing to do with you.

A reference range describes a sample, not a standard of health

Start with how the two numbers beside your result were produced. A laboratory measures a group of men, arranges the results from lowest to highest, and cuts off a slice at each end. What is left in the middle becomes “normal”.

The AUA guideline describes the convention exactly: reference laboratories often define a normal value as ranging within the 5th, or 2.5th, and 95th, or 97.5th, percentiles of a sampled population. Nothing in that procedure asks whether the men in the sample were well. It asks only where they fell relative to each other.

The most careful numbers, and the men they came from

The best answer available to the question “what is normal” came out of a deliberate attempt to fix the problem above. Researchers pooled 9,054 community-dwelling men from four cohort studies — the Framingham Heart Study, the European Male Aging Study, the Osteoporotic Fractures in Men Study, and the Male Sibling Study of Osteoporosis — across the United States and Europe.

A hundred participants from each cohort were re-measured using a reference method at the Centers for Disease Control and Prevention, and the results were used to convert every cohort's figures onto a single common scale. Only then were the percentiles calculated.

The middle figure is the one nobody quotes and the one worth remembering. The median healthy young man in that pooled population sat at 531 ng/dL. A result of 320 is inside the range and also well below the middle of it, and those two facts are both true at once.

Two words in that sentence are doing a great deal of work

“Healthy” and “nonobese” are not decoration. The same paper reports that age-specific concentrations in nonobese men were similar across the four cohorts and greater than in all men taken together. Restricting the sample raised the numbers.

That is the practical difference between a range built to describe health and a range built to describe attendance. The men in a commercial laboratory's sample are, by definition, men who had blood taken — which is not a random slice of the population and is not a healthy one.

It also explains why the same man can be told two different things by two different pieces of paper without either being fraudulent. He is being compared against two different crowds.

Definitive age-specific reference ranges do not exist

This is the sentence that most “levels by age” tables are quietly built on top of. The AUA guideline says it directly: while definitive age-specific reference ranges do not exist, some data suggest that patient age may play a role in setting therapeutic ranges, at least in the elderly population.

It repeats the point in its own list of open research questions, which begins by asking whether there are age-specific reference ranges for testosterone at all, and says the question should be answered for both total and free testosterone using specific methods. When a guideline lists something as future research, that is the honest state of the evidence.

The exception it does cite is instructive. A community-dwelling study of 3,690 healthy men aged 70 to 89 in Perth, Australia established a reference range of 283 to 454 ng/dL, with a mean of 361. Compare that with 264 to 916 in men aged 19 to 39 and the shape of the change becomes visible: the top of the range falls a great deal further than the bottom of it does.

So a chart that hands you a tidy figure for “men in their fifties” is doing something the guideline says cannot yet be done definitively. That does not make the chart malicious. It makes it more confident than its evidence.

What ageing subtracts, measured in the same men over years

There is a better way to ask the age question than comparing different men of different ages, and it was done. The Baltimore Longitudinal Study of Aging measured testosterone in stored samples from 890 men and followed them, so the comparison is within the same man over time rather than between one man and another.

The same study reported how often that adds up to something a clinician would call low. Using total testosterone criteria, the incidence of hypogonadal levels rose to about 20 per cent of men over 60, 30 per cent over 70 and 50 per cent over 80. It also found that levels decreased with increasing body mass index, which is the thread running through this whole article.

Two things follow from that rate of change. The first is that a single year is a small signal, which is why a comparison against an old result is often more informative than a comparison against a printed range. The second is that a large drop over a short period is not the ordinary arithmetic of ageing and deserves a different question.

Why a laboratory can print 250 as normal

The guideline anticipates the confrontation between its own cut-off and the range on your report, and it uses a specific number to do it. A total testosterone value of 250 ng/dL, it says, may be considered low based on the guideline but be marked within the normal range by the laboratory.

Its instruction in that situation is unambiguous. Because the testosterone literature uses absolute values to define low testosterone, the absolute value is the most important factor, and in cases of discrepancy between laboratory reference ranges and the guideline, clinicians are recommended to use the absolute value — with the understanding that all laboratories carry some degree of variability.

This is the single most useful thing on this page. If your result is between roughly 250 and 300 and came back unflagged, you have not been told that the level is fine. You have been told where it sits relative to whoever that laboratory measured. What the cut-off actually means, and what has to accompany it, is set out in our guide to low testosterone after 40.

Two laboratories, two machines, two answers

The other moving part is the measurement itself, and the guideline documents its size rather than glossing over it. Most routine total testosterone is measured by immunoassay, which is fast and high-throughput and has a published intra-assay coefficient of variation of minus 14 to plus 19 per cent.

The important detail is where that error concentrates. The variation is most pronounced at lower testosterone values, reaching 40 per cent in samples below 100 ng/dL — which is precisely the region where the result changes what happens next. The alternative, liquid chromatography with mass spectrometry, is described as the gold standard and must stay within 6.4 per cent to keep its CDC approval.

That tolerance is not a rhetorical figure. The CDC Hormone Standardization Program certifies laboratory methods for total testosterone across a concentration range of 2.50 to 1,000 ng/dL against a reference measurement, and publishes the list of methods that meet it. The harmonisation study reached the same conclusion from the other direction: a substantial proportion of the variation between cohorts was due to assay differences.

What “low” meant across forty studies

If the definition of low were stable, none of the above would matter much in practice. It is not, and there is a measurement of how unstable it is. A review searched the literature from 1966 to 2014 for studies that paired a clinical indication of low testosterone with an actual measurement.

Read that twice. The proportion of men found to be low ranged from one in fifty to three in four, and the largest single driver was where each study drew its own line. The AUA's own response to this mess was to pick one threshold and defend it: total testosterone below 300 ng/dL, chosen with a view to maximising potential benefit while minimising the risks of treatment.

The 300 figure is therefore best understood as a deliberate decision rather than a biological boundary. Nothing changes inside a man's body as he crosses it. What changes is the balance of expected benefit against expected risk from treating him.

How to read your own result against all of this

None of the above means the number is worthless. It means the number needs three pieces of context to be read, and all three are available to you.

1. Find the collection time before the value. An afternoon draw is a different measurement, not a slightly worse one. The testosterone test covers how much difference the hour makes.
2. Compare against the absolute figures, not only the printed band. 300 as the cut-off, 531 as the median of healthy young men, 264 as the bottom of the harmonised range.
3. Check whether there are two results, from the same laboratory. One is a snapshot, and two from different machines are hard to compare.
4. Put the symptoms next to it. The diagnosis requires both halves; which symptoms actually point at it sorts them by how much weight each carries.
5. Ask for the older result if one exists. Given a decline of a few ng/dL a year, your own previous figure is often a more informative comparison than any published range.

That is what turns a printed line into an answer. If the answer comes back as yes, what treatment involves — and what has to be measured before it starts — is set out in testosterone replacement therapy: what each form does. If broken nights are the background to all of this, our sleep guide is the better place to start.