Say your total testosterone comes back at 420, the range printed next to it says 300–1000, and the report says normal. You do not feel normal. Or it is the other way round — the number is under the line, you have no symptoms worth mentioning, and the blood was drawn at four in the afternoon after a bad night's sleep.
Either way you are staring at one number and wondering whether the other number further down the page — the one labelled free — is the one that counts.
Here is what those numbers are, what SHBG has to do with all of them, why free testosterone is usually calculated rather than measured, where that calculation stops being trustworthy, and why the same three numbers are read differently on a woman's report. It is educational, not medical advice. Reading your labs and deciding what to do about them is your prescriber's job.
What is the difference between total and free testosterone?
Total testosterone is everything in the sample; free testosterone is the small fraction that is not stuck to a protein.
Once testosterone is in circulation it is almost all bound. Only about 1–4% travels unbound. Somewhere between a third and a half — 33% to 54% — is held loosely by albumin. The rest is bound tightly by sex hormone-binding globulin, or SHBG.
That gives you three numbers describing the same blood:
- Total T — all of it, bound and unbound.
- Free T — the unbound fraction only.
- Bioavailable T — free plus the albumin-bound portion, which is to say everything not held by SHBG.
The reason anyone bothers is the free hormone hypothesis: the idea that only the unbound fraction can enter tissue and act. It is the model every calculator is built on, but it is still argued over in the literature rather than settled — worth knowing before you treat a free-T figure as the truth and the total as noise.
Why does SHBG change what the same total means?
Because SHBG decides how much of your total is spoken for. Two people with an identical total testosterone can have meaningfully different free fractions if one has high SHBG and the other low.
And SHBG is not a fixed personal constant. It goes up with age, hyperthyroidism, higher estrogen levels including oral estrogen, chronic hepatitis C or HIV, and some anticonvulsants. It goes down with obesity, insulin resistance and type 2 diabetes, hypothyroidism, PCOS, glucocorticoids, exogenous androgens, nephrotic syndrome and acromegaly.
So the relationship between your total and your free is not stable across years, a twenty-pound weight change, or a thyroid diagnosis. That is why a free estimate exists — and why comparing this year's free T to a total from three years ago tells you less than it appears to.
Why is free testosterone calculated instead of measured?
Because the reference method — equilibrium dialysis — is labour-intensive and expensive, and few clinical laboratories run it. What most labs report instead is a number derived from three values with the Vermeulen equation, published in 1999: total testosterone, SHBG and albumin, fed through published binding constants.
That is the same arithmetic behind our free testosterone calculator, which shows the constants it uses rather than handing you a figure from a black box.
An illustrative run of the numbers, using values chosen to demonstrate the math and not to describe anyone: ILLUSTRATIVE — total testosterone 600 ng/dL, SHBG 30 nmol/L, albumin 4.3 g/dL gives free testosterone of roughly 13 ng/dL, about 2% of the total, and bioavailable testosterone of roughly 325 ng/dL. Change nothing but SHBG and the free figure moves while the total sits still.
A third option exists and is the weakest: a free testosterone measured directly by analog immunoassay. Those assays are fast and repeatable but lack accuracy, and have been shown to read free testosterone substantially lower than equilibrium dialysis does. If your report says "free testosterone," find out which of the three you were given.
Where does calculated free testosterone stop being reliable?
At the edges — unusually high or low SHBG, and the female range.
Two specific limits are documented. When SHBG falls below about 15 nmol/L or climbs above about 100 nmol/L, the binding constants the linear equations assume differ substantially from what is actually observed. And when the Vermeulen calculation was checked against direct equilibrium dialysis by mass spectrometry in 183 women and 146 men, it overestimated free testosterone by a median ratio of 1.19 — roughly 19% high — though it was still the most robust of the calculated methods tested.
Read that as a systematic offset, not random error. A calculated free T is useful against your own earlier calculated free T from the same lab and the same three inputs, and much weaker as a single verdict.
Why do two labs give different numbers for the same blood?
Because assays are calibrated differently, and "low" depends partly on whose line you are standing behind. The Endocrine Society's position statement on measuring testosterone argued that a lab should be judged on measuring known concentrations accurately, "not only on agreement with others using the same method." The CDC runs a hormone standardization program that certifies testosterone assays against its own reference measurement procedure, using a ±6.4% mean-bias criterion assessed over four consecutive quarters of blinded samples. An uncertified assay carries no such guarantee.
Reference ranges diverge too. Harmonized data from four cohorts totalling 9,054 men put the range for non-obese men aged 19–39 at 264–916 ng/dL, and the Endocrine Society's 2018 guideline adopts 264 ng/dL as the lower limit for CDC-certified assays. The American Urological Association uses 300 ng/dL as its cut-off. Both are serious bodies; the numbers differ.
The practical consequence is small and annoying: keep repeat draws at the same lab on the same method, which the AUA recommends explicitly, and record the units and the reference range printed on your report rather than one you read online.
Why does the blood draw have to be in the morning?
Because testosterone follows a daily rhythm, and an afternoon draw can look low for reasons that have nothing to do with you. In a study of 66 men sampled at both morning and afternoon visits, men aged 30–40 measured 20–25% lower at 4pm than at 8am. The gap narrowed with age — about 10% by 70 — but the authors still concluded that testosterone measurement should be restricted to morning hours at every age.
The guidelines build that in, and both ask for more than one draw: the Endocrine Society says to measure total testosterone "on two separate mornings when the patient is fasting," and the AUA likewise makes the diagnosis only after two measurements on separate occasions, both early morning. A single low afternoon result is a scheduling artefact until repeated.
What do these numbers mean for women?
The reference range is much lower, and the clinical question is often the opposite one. On a woman's report the guideline-backed use of a testosterone panel is assessing androgen excess — PCOS, hirsutism, irregular cycles — rather than deficiency, for which, as below, no diagnostic blood level exists.
The 2023 international PCOS guideline puts total and free testosterone first-line for assessing biochemical hyperandrogenism, with calculated free T or the free androgen index as acceptable derivations. It also asks laboratories to use mass spectrometry over direct immunoassays, which it describes as having limited accuracy and poor sensitivity and precision at these concentrations.
On the deficiency side, the 2019 Global Consensus Position Statement is direct: no cutoff blood level of any circulating androgen separates women with sexual dysfunction from women without it, research and practice should centre on total testosterone rather than "free" testosterone as the main biomarker, and direct assays for total and free testosterone are "highly unreliable in the female range." Total testosterone by mass spectrometry is what it endorses. The only evidence-based indication it supports is hypoactive sexual desire disorder in postmenopausal women — and it notes that in most countries testosterone for women is prescribed off-label, as a formulation approved for men or a compounded preparation. The point here is narrower: the same lab number means something different on a woman's report.
What to record so the numbers mean something
A testosterone result is almost uninterpretable on its own and quite interpretable in a series. Log, each time:
- Date and time of the draw — the time is not optional, given the morning rhythm above.
- Which lab, and the assay — mass spectrometry, immunoassay, or calculated.
- Total T, SHBG and albumin together — a total without SHBG cannot be turned into anything else later.
- The reference range printed on that report, with units.
- What changed since last time — weight, thyroid status, a new medication, a new estrogen, anything on the SHBG list above.
That last line is what turns a folder of PDFs into something a ten-minute appointment can actually use.
