Total vs free testosterone: why the same blood reads low at one lab and normal at another

By Marcus Renshaw ·

Most people treat a testosterone result the way they treat a cholesterol result: a number, with a range printed beside it, and a verdict implied by where the number sits. It does not work that way. A testosterone result is a number plus the assay that produced it plus the hour it was drawn, and if you get the second or third of those wrong the first one is not so much wrong as unusable. This is the marker where measurement error is large enough to flip the conclusion, and it is the marker where an entire industry is built on the flip.

The same blood sample can read “low” or “normal” depending on the lab and method, because many testosterone assays are not standardized. The fix is two early morning fasting draws on a CDC HoST certified assay, at the same lab. Order total testosterone plus SHBG and albumin if you want a usable free testosterone figure.

Why does the same sample read differently at different labs?

Because a large share of the variation between testosterone results is the assay, not the person. When researchers built harmonized reference ranges from 9,054 men across four cohort studies, they had to cross-calibrate every cohort’s assay against a reference method at the CDC to make the numbers comparable at all. Their conclusion states it plainly: a substantial proportion of the variation between cohorts was due to assay differences rather than biology.

That harmonization is also where the familiar range comes from. In healthy nonobese men aged 19 to 39, the harmonized percentiles run 264 at the 2.5th, 303 at the 5th, 531 at the median, 852 at the 95th and 916 at the 97.5th, in ng/dL. Two things fall out of that list. The commonly used clinical cutoff near 300 ng/dL is essentially the 5th percentile of healthy young men, so it is a screening floor rather than a target. And the difference between normal and optimal is wider here than on almost any other panel item, because the middle of that range is 531.

The CDC runs a certification program for this exact problem. Its Hormone Standardization program for total testosterone, HoST, certifies assays that hold within ±6.4% mean bias against the CDC reference method across 2.50 to 1,000 ng/dL. Certification lasts a year, is earned after four consecutive passing quarters, and is then re-evaluated quarterly. The CDC publishes the list of certified assays and participating laboratories, which means this is one of the rare cases where you can check the instrument behind your own result before you argue with it.

What does the 2026 Endocrine Society statement actually say?

On 16 July 2026 the Endocrine Society published a statement on testosterone replacement therapy that says the quiet part directly: “Non-standardized assays mean the same blood sample can read ‘low’ or ‘normal,’ depending on the lab and method.” Its diagnostic ask is that hypogonadism be diagnosed “based on at least two early-morning, fasting testosterone tests”, with a common clinical threshold near 300 ng/dL, using an assay certified by the CDC HoST program.

One thing to be precise about, because it has been reported both ways. This is a statement, not a new guideline. It restates the Society’s existing Clinical Practice Guideline rather than changing it, and it was issued into a moment of loosening rules and rising prescribing. Nothing in it is a new requirement. That is arguably the more interesting fact: the two draw, early morning, certified assay standard has been the recommendation for years, and the Society felt the need to say it again out loud.

The statement also does not address free testosterone at all. That comes from the guideline underneath it, which matters for getting the attribution right.

How much does the time of day move the number?

Enough to change your diagnosis. In a study that sampled 66 men aged 30 to 80 across six visits, three in the morning and three in the afternoon, testosterone in men aged 30 to 40 ran 20 to 25% lower at 4pm than at 8am. The gap narrows with age, down to about 10% at 70.

The finding worth the price of admission is in the same paper. Seventeen men had at least one afternoon measurement below 300 ng/dL while all three of their morning measurements were normal. That was five of eight men aged 30 to 47 and four of nine aged 66 to 80. An afternoon draw does not shade your result slightly. It can hand you a hypogonadal label that three morning draws would have contradicted.

There is a wrinkle that cuts the other way and is worth knowing before you over-trust the morning rule. A 24 hour sampling study in 21 younger men found the diurnal pattern largely absent in the men who actually had low testosterone: the normal group fell 15% from 8am to 2pm (423 to 358 ng/dL, P = 0.0003), while the low group moved 4% and not significantly (228 to 218 ng/dL, P = 0.54). Two of eleven men in the normal group showed no variation at all. The practical line from that study is the useful one: no participant whose baseline testosterone was above 400 ng/dL dropped below 300 at any point in the day. It is 21 people, so hold it loosely, but it means the men most likely to be mislabeled by a badly timed draw are the ones sitting in the 300s.

Other hormones do not behave this way. The same 66 man study found limited diurnal variation in DHT, SHBG, LH, FSH and estradiol, so those can be drawn through the day without the same care. Testosterone is the one that demands the morning slot.

Total or free testosterone, and which should you order?

Start with total, and add free only when there is a reason. The Endocrine Society’s Clinical Practice Guideline recommends fasting morning total testosterone on an accurate assay as the initial test, confirming it by repeating the morning fasting total, and then obtaining free testosterone in one specific situation: “In men whose total T is near the lower limit of normal or who have a condition that alters sex hormone-binding globulin”.

That second clause is where this audience lives, because plenty of ordinary things move SHBG. The guideline publishes the list. Conditions it associates with increased SHBG are aging, hyperthyroidism, cirrhosis and hepatitis, HIV disease, use of estrogens, some anticonvulsants, and polymorphisms in the SHBG gene. Conditions associated with decreased SHBG are obesity, diabetes mellitus, hypothyroidism, acromegaly, nephrotic syndrome, use of glucocorticoids, some progestins and androgenic steroids, and again SHBG gene polymorphisms. Two entries on that list deserve a self-experimenter’s attention: androgenic steroids lower it, so using testosterone changes the very ratio you are trying to read, and a gene polymorphism can set yours off-centre for life with nothing to find on examination. If SHBG is shifted, total testosterone misreads how much hormone is actually available, in either direction. A man with high SHBG can show a comfortable total and a poor free figure, and a man with low SHBG the reverse. That is the whole case for measuring it rather than assuming it.

Note the guideline says free testosterone by “equilibrium dialysis or estimating it using an accurate formula”, and it says nothing kind about the cheap option, which brings us to the part almost nobody checks.

Why free testosterone numbers are not comparable between methods

There are three ways to get a free testosterone figure and they do not produce the same number.

MethodHow it worksThe catch
Equilibrium dialysisPhysically separates unbound hormone, then measures itReference standard. Slow and expensive, often a send-out
Calculated (Vermeulen)Arithmetic from total testosterone, SHBG and albuminNeeds all three drawn together. Accurate enough, tends to run slightly high
Direct analog immunoassayA single direct measurement, cheapest optionProduces numbers on a different scale entirely

That last row is the trap. In a series where 56 men had split samples run by both methods, the direct radioimmunoassay correlated well with equilibrium dialysis (r = 0.966), and calculated free testosterone correlated slightly better (r = 0.986), but the direct assay’s absolute values came in at roughly one seventh of the dialysis values. Correlation is not agreement. A method can track your changes faithfully and still report a number you cannot compare against a reference range built on a different method, which is exactly how people conclude their free testosterone is catastrophic when it is merely measured differently. That study ran 56 men at a single men’s health clinic, so treat the precise ratio as illustrative rather than a conversion factor, and treat the lesson as firm.

The practical consequence: if you want a free testosterone figure you can track and defend, order total testosterone, SHBG and albumin in the same draw and use the calculated value, or pay for equilibrium dialysis. Do not order the cheap direct assay and then compare it to a number you read somewhere.

What should you actually ask for?

The clinical floor and the optimization version are close together here, which is unusual.

  • Two draws, not one. Both early morning, both fasting. This is the guideline’s own requirement for a diagnosis and it exists because a single value carries too much noise to act on.
  • Before noon, and ideally near 8am. The 20 to 25% afternoon drop is the difference between a real result and a coin flip if your number is anywhere near the 300s.
  • The same lab both times, and a HoST certified assay. Two draws on two different non-standardized assays reproduce the problem you were trying to solve. Check the CDC list.
  • Total testosterone, SHBG and albumin together if a free figure matters to you, which it does if your total sits near the lower limit or anything in your life is moving SHBG.
  • LH and FSH alongside, because they separate a testicular problem from a pituitary one and they do not care what time you draw them.
  • Record the assay and the lab with the number, not just the number. A result without its method is not a data point you can compare to next year’s.

The budget logic is worth stating because it cuts against the instinct to buy the big panel. Two properly timed total testosterone draws on a certified assay at one lab are worth more than a twenty marker hormone panel drawn once at 3pm at whichever lab was cheapest. Retest cadence follows the marker. Six weeks is the figure worth knowing, though it is worth being precise about where it comes from: it is set by the half-life of an injectable testosterone ester reaching a steady level, so it applies cleanly to a dose change on testosterone itself. For anything else that moves the number, a SERM, weight loss, better sleep, treating a thyroid problem, the settling time is not as well characterised, and six weeks is a reasonable floor rather than a rule. Either way a draw taken before then buys noise at full price.

What if you are already running something?

Then the measurement problem compounds, and this is where the guideline stops being enough. Clinical guidance is written for a man being assessed once for a diagnosis. If you are on a protocol, you are asking a different question: not “am I hypogonadal” but “did what I changed move the number, and by how much”.

That reframes every rule above into a consistency requirement. Same assay, same lab, same hour, same fasted state, every time, because you are now measuring a difference rather than a level, and a method change between two draws will produce a difference that is entirely artifact. Several things in a typical stack move these markers directly: testosterone suppresses LH and FSH and pushes hematocrit and estradiol, a SERM like enclomiphene moves LH, FSH and estradiol upward, and weight loss on a GLP-1 changes SHBG and with it the relationship between your total and free figures. If you are running more than one, the result is a net effect and not a clean signal, which is an argument for changing one variable at a time whenever the calendar allows it.

My own total testosterone sat mid-range and was flagged normal, and I still chose to move it, which is how I ended up on enclomiphene rather than testosterone itself. That decision was made on a total, and the honest criticism of deciding on a total is that without SHBG beside it you do not know whether the available fraction agrees with the number you are looking at. The part I would defend is keeping the draws consistent and changing one thing at a time. The part I would tell anyone starting from scratch is to spend the extra on SHBG before spending it on anything exotic.

If your last testosterone result came from an afternoon draw, at a lab you did not choose, on an assay you cannot name, then the number you have been reasoning from is not really a measurement of you. That is a cheap thing to fix, and it is worth fixing before you act on it, not after. None of this is medical advice, and a hormone decision belongs in a conversation with a clinician who can see the whole panel.

FAQ

Should I test total or free testosterone?

Start with total testosterone, drawn fasting in the early morning on a certified assay, and repeat it to confirm. Add free testosterone when your total sits near the lower limit of normal or when something is altering your sex hormone binding globulin, which includes obesity, insulin resistance, thyroid disease, liver disease and androgen use. That is the Endocrine Society guideline’s own trigger for adding it rather than ordering it by default.

Why do two labs give different testosterone results for the same blood?

Because many testosterone assays are not standardized against a common reference method, so they carry different biases. The Endocrine Society put it directly in July 2026: non-standardized assays mean the same sample can read low or normal depending on the lab and method. The CDC HoST program certifies assays that stay within 6.4% mean bias of its reference method, and publishes the list of which ones qualify.

What time of day should I get testosterone tested?

Early morning, fasting, ideally close to 8am. In men aged 30 to 40, levels run 20 to 25% lower at 4pm than at 8am, and the gap shrinks to about 10% by age 70. In one study, 17 men had an afternoon reading below 300 ng/dL while every one of their morning readings was normal.

Is one testosterone test enough?

No. The diagnostic standard is at least two early morning fasting measurements, because a single value carries too much biological and analytical noise to support a decision. If the two disagree meaningfully, that is information too, and usually a sign to check the timing and the assay before concluding anything about yourself.

What is a normal testosterone level?

The harmonized range for healthy nonobese men aged 19 to 39 is 264 to 916 ng/dL, standardized against a CDC reference method. Worth knowing where the landmarks sit: 303 is the 5th percentile and 531 is the median. The commonly used clinical cutoff near 300 is therefore close to the bottom 5% of healthy young men, which makes it a screening floor rather than a goal.

Why is my free testosterone number so low compared to the reference range?

Possibly because it was measured by a method the range was not built for. Direct analog immunoassays report values on a different scale from equilibrium dialysis, roughly a seventh of them in one split sample series, while still correlating well. If you want a free figure you can compare and track, use equilibrium dialysis or the calculated value from total testosterone, SHBG and albumin drawn together.

Does SHBG really change the picture that much?

It can invert it. Sex hormone binding globulin determines how much of your total testosterone is actually available, so a man with high SHBG can show a reassuring total and a poor free level, and a man with low SHBG the opposite. Per the Endocrine Society guideline’s own table, SHBG rises with aging, hyperthyroidism, liver disease and estrogen use, and falls with obesity, diabetes, hypothyroidism and androgenic steroid use, with gene polymorphisms able to shift it either way. Those are common enough that assuming yours is normal is a guess rather than a finding.

How long after starting or changing something should I retest?

About six weeks is the usual answer, and it is firmest for a dose change on injectable testosterone, where the interval is set by the ester’s half-life reaching a steady level. For other things that move testosterone the settling time is less well characterised, so treat six weeks as a floor rather than a rule. Either way an early draw shows you a number still in transit. Keep every variable of the measurement itself identical between draws, same lab, same assay, same early morning fasted slot, since you are measuring a change and a method difference would masquerade as one.

Sources