Most men meet their endocrine system as a PDF.
What each test on a male hormone panel actually measures, what moves it, and why a reference range is not a target.
Most men meet their endocrine system as a PDF.
A row of analytes, a value, a reference range, and a flag in the margin. The flag is doing an enormous amount of work — it is the thing that turns a number into a worry, or into a prescription. This section exists to take the flag less seriously and the number more seriously.
Each page covers one analyte: what it measures, what raises and lowers it, when it is worth ordering, how it is misread, and what it means before and during testosterone treatment.
The pages
Total testosterone — the number the whole evaluation rests on, and the one most changed by the clock, the last meal and the assay.
Free testosterone — three different tests reported under one name, on scales that differ by up to tenfold.
SHBG — why total and free testosterone disagree, and why lowering it is not a treatment.
LH — the test that tells you whether the problem is in the testis or above it.
FSH — what it says about sperm production, and the claim about sperm retrieval that runs backwards.
Estradiol — the floor is real and well evidenced; the ceiling everyone titrates to is not.
Prolactin — the test that finds the pituitary tumours, and the one with the most ways to be wrong.
PSA — what testosterone does to it, and where the referral thresholds actually came from.
Hematocrit — the test most likely to change a testosterone prescription.
DHT — the most potent androgen, and the one no guideline asks you to measure.
TSH and free T4 — the test that can reverse the meaning of a testosterone result.
Inhibin B — good at grading sperm production, and asked to do two jobs it cannot do.
HbA1c, glucose and lipids — where the observational literature and the randomised literature tell two different stories.
Read these in groups rather than one at a time. Total testosterone, free testosterone and SHBG are a single conceptual unit; so are LH, FSH and inhibin B. Hemoglobin travels with hematocrit and adds nothing separate, and albumin matters here only as an input to a calculated free testosterone — both are covered on those pages rather than given their own.
Two rules that apply to every page here
A reference range is not a target. It is the middle 95% of whatever population the laboratory used to build it. A value inside the range is not automatically fine, and a value outside it is not automatically a problem. The range answers “is this unusual?” It does not answer “is this a problem for you?”, and those are different questions with different consequences.
We do not publish a single universal target for anything. Not for testosterone, not for estradiol, not for hematocrit. Where a society has stated a number we quote it and attribute it. Where societies disagree we show the disagreement rather than picking the tidiest one. And where a number is quoted everywhere but traceable nowhere, we say that too — a table that looks authoritative and isn’t will be screenshotted and quoted back for years.
Why one abnormal value rarely means anything on its own
Three reasons, and they compound.
Biological variation. Hormones are not constants. In men sampled repeatedly over a year, the within-person coefficient of variation was about 11% for total testosterone and 12% for SHBG. For LH it is far worse — 22% in one study of weekly sampling over ten weeks, 26% in an independent replication. A single LH is a weak basis for classifying a borderline case, and that is a fact about the hormone, not about the laboratory.
Assay variation. Two laboratories measuring the same serum can disagree meaningfully. In a comparison of six immunoassays against mass spectrometry in men, one platform ran about 90 ng/dL low on average and another about 99 ng/dL high. That spread is wider than the gap between several published diagnostic thresholds.
Context. Testosterone falls after a meal — about 25% after a 75-gram glucose load, and about 30% after an ordinary 550-calorie meal — with no change in LH or SHBG, so the fall is real rather than an artefact of binding. It falls further and faster in acute illness: within a day of a heart attack, a head injury or elective surgery, mean testosterone dropped by roughly 200 to 270 ng/dL in one serial study, recovering as the men recovered. A hormone panel drawn in the wrong week describes the week, not the man.
How to read your own report
Start with the question the panel was meant to answer. If it was “is my testosterone low”, then the relevant analytes are total testosterone measured correctly, SHBG and free testosterone where binding is in doubt, and LH and FSH to locate the problem. Everything else on a direct-to-consumer panel is there because it was cheap to add.
Then ask what would change if the value were different. A number that would not alter a decision is not worth acting on, however far outside the range it sits. This is the single most useful filter a patient can apply to their own report, and it is the one commercial panels are designed to defeat.
For the full account of how testosterone itself should be measured — timing, fasting, repetition, assay — start at how testosterone should actually be tested.
Questions patients ask
My result was flagged, so something is wrong.
A flag means unusual for that laboratory's reference population. It does not mean wrong.
What the evidence showsA reference interval is conventionally the central 95% of a reference population, so by construction 1 in 20 healthy people fall outside it on any single analyte. Order twenty analytes on a healthy man and a flag somewhere is the expected result, not the exception.
What remains uncertainNothing about the arithmetic.
Bottom lineThe flag starts the conversation. It does not finish it.
Strong
If my level is in range, my symptoms can't be hormonal.
Not necessarily — but the alternative explanations deserve a hearing first.
What the evidence showsReference ranges are built from populations, not from individuals, and the within-person variation in testosterone is smaller than the between-person variation. A man whose usual value sits high in the range can fall substantially and still be "in range." That is a real phenomenon. It is also the exact argument used to justify treating men who do not need it, so it needs evidence in the individual case rather than assertion.
What remains uncertainWhether within-individual decline predicts symptom benefit from treatment. No trial has tested it.
Bottom linePossible, unproven, and easily abused. Ask what else was ruled out.
Limited
More tests give a clearer picture.
Past a point, more tests give more flags and less clarity.
What the evidence showsEvery added analyte carries its own false-positive rate, and analytes measured without an indication have no established action attached to them. The guideline panels for evaluating low testosterone are short for this reason.
What remains uncertainNothing that would justify a broad screening panel.
Bottom lineOrder what will change a decision.
Strong
I should get the same panel repeated every few months to track trends.
Only if you are on treatment, and then only on the schedule that matters.
What the evidence showsFor a man not on testosterone, repeated broad panels mostly measure biological noise — the within-person variation above means a "trend" across two or three points is usually not a trend at all. For a man on treatment, the schedule is defined: testosterone, hemoglobin and hematocrit at 3 to 6 months, then annually, with PSA where prostate cancer detection would change management.
What remains uncertainThe optimal long-term monitoring interval beyond the first year.
Bottom lineMonitoring on treatment, yes. Surveillance off treatment, rarely.
Moderate
Ready for testosterone care built on a diagnosis?
The first step is a full endocrine evaluation, not a prescription. We see patients across San Diego County and welcome referrals from other physicians.
