Free Testosterone Calculator: Five Published Methods and Reference Ranges Compared
By TRT Provider Guide · Last verified: September 2, 2026 · Research and reference resource
The number worth quoting: Labcorp’s current adult male reference intervals for direct and calculated free testosterone have zero numerical overlap. The highest direct-method upper limit is 26.5 pg/mL. The lowest calculated-method lower limit is 30.3 pg/mL. (Labcorp direct method; Labcorp calculated method)
That does not show that one method is “wrong.” It shows why a free testosterone number must stay attached to the method and reference interval used by the lab.
Figure 1. Same lab, same units, zero numerical overlap. Labcorp’s direct and calculated free testosterone reference intervals are separated in all five adult age-group comparisons. The age bands are close, not identical, and the two methods are not interchangeable.
Source: Labcorp test 144980, Testosterone, Free, Direct and Labcorp test 140226, Testosterone Free, Profile I. TRT Provider Guide comparison, verified September 2, 2026. Chart data CSV.
Free Testosterone Calculator Results at a Glance
What the result means: This tool compares arithmetic. It does not diagnose low testosterone and it does not tell you to start, stop, or change treatment. Read the number with the method, units, reference interval, symptoms, and repeat testing used by your clinician. (Clinical guideline)
The bottom line
A free testosterone calculator gives you one number. That number is not one number.
We collected seven published free testosterone reference ranges that apply to men in their thirties. For men in their thirties, the bottom of "normal" runs from 8.7 pg/mL to 120 pg/mL across seven published ranges — a 13.8-fold spread. The Endocrine Society's own clinical practice guideline says why: A harmonized reference range for FT has not been established.
So the honest answer to "is my free testosterone low?" is: it depends which range your lab used, and nobody has agreed on one.
Here is the part that surprised us. Two of those seven ranges come from studies published in the last three years. Both measured free testosterone directly, using the same reference-method class — equilibrium dialysis followed by mass spectrometry — in healthy, nonobese young men. Their bottom-of-normal numbers are 2.26 times apart.
The calculator below runs your numbers through five published equations at once and shows you what all seven ranges would say. Then we show our work.
[CALCULATOR — see free-testosterone-calculator.html. Placement: directly here, above every other section. Three inputs: total testosterone, SHBG, albumin. No result badge, no verdict, no sign-up.]
What the calculator does: it takes your total testosterone, your SHBG, and your albumin, and estimates the small slice of testosterone that is floating free. It runs five published equations side by side. It shows the answer in pg/mL, ng/dL, and as a percentage. It does not diagnose anything, it does not store what you type, and it does not send you anywhere.
What is free testosterone, in plain words?
Answer capsule: Most of the testosterone in your blood is stuck to proteins and cannot be used. Free testosterone is the small slice that is not stuck to anything — usually around 1% to 3% of your total. Because it is such a small slice, it is hard to measure, and most labs estimate it with a formula instead.
Think of your bloodstream as a bus.
Most of your testosterone is strapped into a seat. It is held tight by a protein called SHBG — sex hormone-binding globulin. Held that tight, it cannot get off the bus and do anything.
Some testosterone is holding a strap, loosely, next to a protein called albumin. It can let go pretty easily.
And a small number of passengers are standing by the open door. Nothing is holding them. That is your free testosterone.
Labcorp puts the size of that group at approximately 0.5% to 2% of your total on one of its test pages. Other sources say 1% to 3%. The exact figure moves around a little depending on who you ask, which is itself a small hint about how this whole area works.
The reason anyone cares: the free slice is the part that can actually get off the bus and reach your tissues. Two men can have the exact same total testosterone and very different free testosterone, because one of them has much more SHBG holding it down.
Terms you will see on this page
| Term | What it means |
|---|---|
| Total testosterone | All the testosterone in your blood, stuck and unstuck together |
| SHBG | The protein that holds testosterone tightly. High SHBG means less free testosterone |
| Albumin | A protein that holds testosterone loosely |
| Free testosterone | The slice held by nothing |
| Bioavailable testosterone | The free slice plus the loosely-held slice |
| Calculated free testosterone | An estimate from a formula, using total T, SHBG, and albumin |
| Measured free testosterone | An actual lab measurement, not an estimate |
| Equilibrium dialysis | The slow, expensive lab method treated as the gold standard for measuring it |
| Direct analog immunoassay | A fast, cheap lab method that estimates it a different way |
| pg/mL and ng/dL | Two units for the same thing. 10 pg/mL = 1 ng/dL |
Source: Labcorp direct free testosterone and Labcorp calculated free testosterone.
Is there one normal free testosterone range?
Answer capsule: No. As of September 2, 2026, there is no single agreed normal range for free testosterone in men. We found seven published ranges that apply to men in their thirties, and their lower limits run from 8.7 pg/mL to 120 pg/mL. The Endocrine Society's 2018 guideline states directly that a harmonized reference range has not been established.
This is the core of it, so here are the seven ranges in one table, in one unit.
| Source | How the number was produced | Ages | Low | High |
|---|---|---|---|---|
| Labcorp test 144980 | Direct analog immunoassay | 30–39 | 8.7 | 25.1 |
| Quest, published by Michigan Medicine MLabs | Referral-lab free testosterone range | 18–69 | 35.0 | 155.0 |
| Labcorp test 140226 | Calculated, Vermeulen equation | 31–40 | 42.3 | 190.0 |
| Quest, published in Endotext Table 4 | Equilibrium dialysis, 5th–95th percentile | 30–39 | 45.0 | 144.0 |
| Walravens et al. 2026 | Equilibrium dialysis with LC–MS/MS | 18–39 | 53.1 | 216.1 |
| Ho et al. 2006 | Calculated, Vermeulen equation, 2.5th–97.5th percentile | 20–45 | 70.7 | 226.4 |
| Jasuja et al. 2023 | Standardized equilibrium dialysis, 2.5th–97.5th percentile | 19–39 | 120.0 | 368.0 |
Source: TRT Provider Guide analysis of Labcorp test 144980, Labcorp test 140226, Michigan Medicine MLabs, Endotext Table 4, Ho et al. 2006, Jasuja et al. 2023, and Walravens et al. 2026. Verified September 2, 2026.
Figure 2. Seven published ranges, one unit. The lower limit runs from 8.7 to 120 pg/mL, a 13.8-fold spread. The methods, populations, percentiles, and age bands differ, so these intervals are not interchangeable.
Source: the seven sources listed under Table 1. TRT Provider Guide normalization to pg/mL, verified September 2, 2026. Chart data CSV.
The numbers that come out of that table:
- Lower limits: 8.7 to 120.0 pg/mL — a 13.8-fold spread
- Upper limits: 25.1 to 368.0 pg/mL — a 14.7-fold spread
- Remove the direct analog immunoassay row, and six ranges remain. Their lower limits still run 35.0 to 120.0 pg/mL — a 3.4-fold spread
We want to be fair here, so we ran that last check on purpose. Even after removing the direct analog method that the Endocrine Society guideline says should not be used to measure free testosterone, the lower endpoint still more than triples depending on which source you pick.
The two studies that should agree, and don't
This is the finding we did not expect.
Jasuja et al. 2023 measured free testosterone in 145 healthy, nonobese men using a standardized equilibrium dialysis method with a CDC-certified mass spectrometry assay. For men aged 19–39, they reported a range of 120 to 368 pg/mL.
Walravens et al. 2026 measured free testosterone by equilibrium dialysis followed by LC–MS/MS in a cohort drawn from 1,202 White men. For healthy, nonobese men aged 18–39, the reported interval of 184 to 749 pmol/L converts to about 53.1 to 216.1 pg/mL.
Same measurement-method class. Nearly the same age group. Both studies restricted their young reference groups to healthy, nonobese men. The studies still produced different intervals.
| Study | Ages | Range (pg/mL) | Range (pmol/L) |
|---|---|---|---|
| Walravens et al. 2026 | 18–39 | 53.1–216.1 | 184–749 |
| Jasuja et al. 2023 | 19–39 | 120.0–368.0 | 415–1,274 |
| Difference | 2.26× at the low end; 1.70× at the high end |
Source: Jasuja et al., Andrology 2023 and Walravens et al., Journal of Clinical Endocrinology & Metabolism 2026. Conversion used for the displayed values: pg/mL = pmol/L × 0.28842. Verified September 2, 2026.
A man aged 30 with a measured free testosterone of 100 pg/mL falls inside the 2026 study interval and below the 2023 study interval. Nothing about him changed. Only the paper changed.
The likely reasons are real and boring: different groups of men, different countries, different dialysis setups, different statistics. That is exactly the point. Reference ranges are not facts about biology. They are facts about a specific group of people measured a specific way.
What this data shows, and what it does not
Answer capsule: This page shows that published free testosterone ranges disagree by a wide margin and that different equations produce different answers from the same blood. It does not show that any one lab is wrong, that your own result is wrong, or that you should ignore what your doctor tells you.
We want to be very clear about the limits of what we found, because overstating it would make it useless.
What this data does show
- Published reference ranges for free testosterone in men disagree substantially, and the disagreement is documented in each source's own paperwork.
- The disagreement comes from two things at once: the method used to produce the number, and the group of men used to define normal. Neither is printed on your lab report.
- Published equations for calculating free testosterone produce different answers from identical inputs. This is arithmetic, and it is reproducible.
- A calculated number and a measured number are different things and should not be read against each other's ranges.
What this data does not show
- It does not show any lab is doing something wrong. Labcorp states plainly on its own test page that its direct assay reads lower than dialysis. The labs are not hiding this. The problem is at the interpretation step, where a patient sees two numbers in the same units and reasonably assumes they mean the same thing.
- It does not show which range is correct. We are not in a position to say, and neither is anyone else right now. That is what "no harmonized range" means.
- It does not tell you whether you are deficient. Free testosterone is one input among symptoms, repeat morning testing, and full clinical history.
- Our model grid is not patient data. Where we run scenarios through equations, those are calculations, not measurements from real men. We label them that way every time.
How we built this
Answer capsule: We implemented five published free testosterone equations from their original papers, then checked our code against two datasets that were published before ours. Our engine reproduced the ISSAM worked example to six decimal places and matched a 126-man published cohort within 1.23% on every equation. Only then did we compute anything new.
We think you should not trust a calculator that will not show its work. So here is ours.
Step 1: we implemented the equations from the source papers
We coded five published methods, using each paper's own binding constants:
| Equation | Year | Type | Key constants | Uses albumin? |
|---|---|---|---|---|
| Vermeulen | 1999 | Law of mass action | SHBG 1.0×10⁹, albumin 3.6×10⁴ L/mol; albumin MW 69,000 | Yes |
| Vermeulen, alternate albumin mass | 1999 | Law of mass action | Same constants; albumin MW 66,500 | Yes |
| Södergård | 1982 | Law of mass action | SHBG 0.6×10⁹, albumin 4.06×10⁴ L/mol | Yes |
| Nanjee–Wheeler | 1985 | Empirical, from gel filtration | Regression on total T and SHBG | No |
| Ly–Handelsman | 2005 | Empirical, from ultrafiltration | Regression fitted on 4,000+ samples | No |
The Ly–Handelsman method is piecewise. For total testosterone below 5 nmol/L, the calculator must use the paper’s low-testosterone equation. At 5 nmol/L or above, it must use the second equation. Applying only the second equation to every input gives wrong low-input results. (Ly–Handelsman primary paper)
Source: equations as written out in full by Ho et al., Ann Clin Biochem 2006;43:389–97, using constants from Vermeulen et al., J Clin Endocrinol Metab 1999;84:3666–72; Södergård et al., J Steroid Biochem 1982;16:801–10; Nanjee & Wheeler, Ann Clin Biochem 1985;22:387–90; Ly & Handelsman, Eur J Endocrinol 2005;152:471–8.
Step 2: we made the code prove itself before using it
A calculator that has never been checked against a published answer is just a guess with a nice font. We ran two checks.
Check one — the ISSAM worked example. The International Society for the Study of the Aging Male publishes a worked example with its calculator: total testosterone 288.4 ng/dL, SHBG 40 nmol/L, albumin 4.3 g/dL. Their published answers are 5.02 ng/dL free and 118 ng/dL bioavailable.
| Output | ISSAM published | Our engine | Second independent implementation |
|---|---|---|---|
| Free testosterone | 5.02 ng/dL | 5.014772 ng/dL | 5.014805 ng/dL |
| Percent free | 1.74% | 1.74% | — |
| Bioavailable testosterone | 118 ng/dL | 117.520091 ng/dL | 117.520875 ng/dL |
Source: ISSAM worked example, issam.ch/freetesto.htm. TRT Provider Guide calculation, September 2, 2026. The second implementation was written separately, in a different programming language, from the same source papers; the two agree to within 0.0007%, a gap consistent with a slightly different value for testosterone's molecular weight.
Check two — a published 126-man cohort. Ho et al. reported average results for four equations across 126 healthy men aged 20–45. We fed our engine their published cohort averages and compared.
| Equation | Published mean (nmol/L) | Our engine (nmol/L) | Difference |
|---|---|---|---|
| Vermeulen | 0.442 | 0.4385 | −0.79% |
| Södergård | 0.465 | 0.4653 | +0.07% |
| Nanjee–Wheeler | 0.524 | 0.5220 | −0.38% |
| Ly–Handelsman | 0.343 | 0.3472 | +1.23% |
Source: TRT Provider Guide calculation using cohort means published by Ho et al., Ann Clin Biochem 2006;43:389–97 (total testosterone 20.2 nmol/L, SHBG 30.3 nmol/L, albumin 46 g/L). Verified September 2, 2026.
Every equation landed within 1.23% of its published value, and three of four within 0.8%. That is the standard we held ourselves to before publishing a single new number.
Step 3: we built the comparison grid
We ran six total testosterone levels (250 to 600 ng/dL) against eight SHBG levels (10 to 80 nmol/L), with albumin fixed at 4.3 g/dL, through all five equations. That is 48 scenarios and 240 calculated values.
These are modeled scenarios, not patients. No real person's blood is in that grid. It exists to show how the equations behave, not to describe any population.
Step 4: we kept the CDC assay comparison separate
The CDC abstract reports bias from testing the same serum samples by different methods. Our Labcorp comparison uses endpoints from two published reference-interval tables. Those are different analyses, so one cannot validate the other. We report both, but we do not treat similar ratios as proof that they measure the same thing.
Why did two labs give me different free testosterone results?
Answer capsule: Because "free testosterone" is the name of at least three different lab products. Labcorp alone sells two tests with that name, both reported in pg/mL, and their adult male reference ranges do not overlap at all in any of the five comparable age bands.
This is the most concrete thing on this page, and you can check it yourself in about two minutes.
Labcorp sells test 144980, "Testosterone, Free, Direct." It uses a direct analog immunoassay. Results come back in pg/mL.
Labcorp also sells test 140226, "Testosterone Free, Profile I." One of its listed synonyms is literally "Free Testosterone by Vermeulen Equation." It is calculated, not measured. Results come back in pg/mL.
Same lab. Same units. On Labcorp's own test pages, both results map to the same LOINC code, 2991-8, which is the standard code labs and medical records use to identify a test. And here are the ranges. The LOINC record leaves the method field unspecified, so the shared code does not make the direct and calculated methods interchangeable.
| Age | Test 144980 (measured) | Test 140226 (calculated) | Overlap | Gap between them | Ratio of lower limits |
|---|---|---|---|---|---|
| 20s | 9.3 – 26.5 | 47.7 – 173.9 | None | 21.2 pg/mL | 5.13× |
| 30s | 8.7 – 25.1 | 42.3 – 190.0 | None | 17.2 pg/mL | 4.86× |
| 40s | 6.8 – 21.5 | 30.3 – 183.2 | None | 8.8 pg/mL | 4.46× |
| 50s | 7.2 – 24.0 | 35.8 – 168.2 | None | 11.8 pg/mL | 4.97× |
| 60s | 6.6 – 18.1 | 34.7 – 150.3 | None | 16.6 pg/mL | 5.26× |
Source: TRT Provider Guide analysis of reference range tables published on labcorp.com for test 144980 and test 140226. Age bands are closely matched, not identical (144980 uses 20–29, 140226 uses 18–30, and so on). Verified September 2, 2026.
Zero overlap in five out of five age bands.
Here is what that means for a real person. A man in his twenties gets a free testosterone result of 37 pg/mL.
- On Labcorp test 144980, the interval tops out at 26.5. His result is above that published interval.
- On Labcorp test 140226, the interval starts at 47.7. His result is below that published interval.
One lab. One number. Two opposite answers. We call the space between the two ranges the dead zone, and for men in their twenties it is 21.2 pg/mL wide.
To be fair to the labs
Labcorp is not hiding any of this. Its own test page for the direct assay states that in the original comparison work the analogue results were only about one-fourth as high as the dialysis method. The lab knows. The lab published it.
The gap opens up at the interpretation step. A patient sees two numbers in the same units under the same name and assumes they are comparable. Nothing on a typical lab report tells him otherwise.
The CDC measured this in 2026
In June 2026, the CDC's Clinical Standardization Programs presented an interlaboratory comparison at ENDO 2026. It is worth reading carefully.
They tested 10 free testosterone methods across 8 laboratories and assay manufacturers: six lab-developed dialysis methods, two commercial immunoassays, and two calculated methods based on the Vermeulen formula. They ran 43 single-donor serum samples, in duplicate, on two separate days. Results ranged from 2.9 to 339 pg/mL. Everything was compared against the CDC's own dialysis mass-spectrometry method.
| Method group | Mean bias vs the CDC reference method |
|---|---|
| Direct immunoassays | −86.5% to −67.7% |
| Some dialysis lab-developed tests | −9.2% to +2.1% |
| Other dialysis lab-developed tests and calculated methods, grouped | +46% to +136% |
Source: Zhou H, Ribera A, Sugahara O, Danilenko U, Vesper H. "Assessment of Free Testosterone Measurement Methods: Findings from an Interlaboratory Comparison Study." CDC Clinical Standardization Programs. ENDO 2026 abstract SUN-144, presented June 14, 2026. Note: this is a conference abstract, not yet a peer-reviewed paper. The abstract groups calculated methods together with some dialysis tests in the +46% to +136% figure, so that range should not be read as applying to calculated free testosterone alone.
The CDC's conclusion was that the field needs standardization, and that the findings support the existing recommendation against using direct immunoassays in clinical practice.
The CDC bias figures and the Labcorp reference-interval gaps answer different questions. The CDC compared methods on the same serum samples. Table 6 compares endpoints from two separate published reference intervals. Their ratios should not be treated as a validation of each other.
How accurate is calculated free testosterone?
Answer capsule: In the largest published head-to-head we found, the Vermeulen calculation ran a median 1.19 times higher than directly measured free testosterone, with 95% of results falling between 0.90 and 1.47 times the measured value. That is a real bias, but it varies enough person to person that you cannot simply subtract 19% from your own result.
Fiers and colleagues analyzed serum from 183 women and 146 men, measured free testosterone directly by dialysis with mass spectrometry, and then compared three calculation methods against it.
| Calculation method | Median ratio to measured | 2.5th–97.5th percentile | What it means |
|---|---|---|---|
| Vermeulen | 1.19 | 0.90 – 1.47 | Runs about 19% high on average |
| Ly–Handelsman | 1.00 | 0.69 – 1.42 | Right on average, but scatters more widely |
| Zakharov | 2.05 | 1.26 – 3.26 | Runs about twice as high |
Source: Fiers T, Wu F, Moghetti P, Vanderschueren D, Lapauw B, Kaufman JM. "Reassessing Free-Testosterone Calculation by Liquid Chromatography–Tandem Mass Spectrometry Direct Equilibrium Dialysis." J Clin Endocrinol Metab 2018;103(6):2167–74 (PMID 29618085). Ratios are from European Male Aging Study samples.
Three things worth pulling out of that table.
One. The Vermeulen equation — the one nearly every calculator uses, including ours — reads about 19% high on average against direct measurement. But the spread runs from 0.90 to 1.47. That is a 1.63-fold range across the middle 95% of people. You cannot correct your own number by taking 19% off, because you do not know where in that spread you fall.
Two. The Zakharov method is a newer, more sophisticated model. Its authors reported that it matched dialysis closely. In this independent dataset it ran a median 2.05 times higher than measured. Two published methods, applied to the same serum, 1.72 times apart from each other. We are not taking a side on which model is better biologically; we are reporting that the published evidence disagrees, which is why we did not implement Zakharov in our calculator. We do not publish numbers we have not derived and validated ourselves.
Three. In those men, the median directly measured free testosterone was 248 pmol/L, or about 71.5 pg/mL, which worked out to 1.5% of their total testosterone.
Which equation should a free testosterone calculator use?
Answer capsule: Most calculators use the Vermeulen 1999 equation, and that is a defensible default. But across 48 modeled scenarios, the five published equations disagreed by a median of 78%. Even limiting the comparison to the three equations that model the same physics, the median disagreement was 14.6%.
Here is what happens when you feed one blood panel to five equations.
| SHBG (nmol/L) | Vermeulen | Vermeulen, alt. albumin | Södergård | Nanjee–Wheeler | Ly–Handelsman | Spread |
|---|---|---|---|---|---|---|
| 10 | 132.0 | 128.2 | 124.5 | 149.2 | 74.0 | 101.6% |
| 20 | 105.0 | 102.5 | 107.6 | 120.5 | 68.9 | 75.1% |
| 30 | 85.9 | 84.2 | 94.3 | 103.8 | 63.7 | 62.9% |
| 40 | 72.2 | 70.9 | 83.7 | 91.9 | 58.5 | 57.0% |
| 60 | 54.1 | 53.4 | 67.9 | 75.1 | 48.2 | 55.8% |
| 80 | 43.0 | 42.6 | 56.9 | 63.2 | 37.9 | 66.8% |
Source: TRT Provider Guide model grid, calculated September 2, 2026, using published equations and constants. All values in pg/mL. These are calculated scenarios, not measurements from real men.
Figure 3. The same inputs, five calculated answers. At total testosterone 400 ng/dL, albumin 4.3 g/dL, and SHBG 40 nmol/L, the five implemented calculations return 58.5 to 91.9 pg/mL.
Source: TRT Provider Guide model grid calculated from the published equations. These are modeled scenarios, not patient data. Calculated September 2, 2026. Chart data CSV.
Across the full 48-scenario grid:
- Median disagreement between the highest and lowest equation: 78.2%
- Range of disagreement: 34.8% to 212.0%
- Scenarios where the highest equation was at least 50% above the lowest: 40 of 48
That headline number is inflated by the two empirical equations at the edges of the data they were fitted on, so here is the stricter version. Limit it to the three equations that all model the same physics — same law of mass action, same idea, only the published constants differ:
- Median disagreement: 14.6%
- Maximum: 35.5%
That is the honest apples-to-apples figure, and it is the one we would want quoted.
Vermeulen and Södergård swap places at SHBG 17
One detail that surprised us. These two equations do not just differ by a fixed amount. They cross over.
| SHBG (nmol/L) | Vermeulen (pg/mL) | Södergård (pg/mL) | Södergård vs Vermeulen |
|---|---|---|---|
| 10 | 114.4 | 108.5 | −5.2% |
| 15 | 101.3 | 100.6 | −0.7% |
| 20 | 90.5 | 93.6 | +3.4% |
| 30 | 74.0 | 81.9 | +10.7% |
| 40 | 62.1 | 72.6 | +16.9% |
| 60 | 46.6 | 58.9 | +26.4% |
| 80 | 37.1 | 49.4 | +33.2% |
Source: TRT Provider Guide model grid, calculated September 2, 2026. Calculated scenarios, not patient data.
Below about SHBG 17, Södergård reads lower than Vermeulen. Above it, Södergård reads higher, reaching 33% higher at SHBG 80. So which equation "reads high" is not a fixed fact — it depends on your SHBG.
Even "the Vermeulen equation" is not one equation
Here is a small thing with a real effect. The Vermeulen equation needs albumin converted from grams into moles, which requires albumin's molecular weight. Two values are in circulation.
Same equation name. Same binding constants. Different albumin mass. In our grid, that alone changes the answer by 0.78% to 3.12%.
It is a small effect. We mention it because most calculators do not tell you which one they used, and it means two sites can both honestly say "we use the Vermeulen equation" and return different numbers.
How much does SHBG change your free testosterone?
Answer capsule: SHBG is the single biggest driver of the result. In our model grid, holding total testosterone at 500 ng/dL and albumin at 4.3 g/dL, calculated free testosterone fell from 168.0 pg/mL at SHBG 10 to 45.5 pg/mL at SHBG 100 — a 72.9% drop with no change in total testosterone at all.
| SHBG (nmol/L) | Calculated free T (pg/mL) | Change from SHBG 10 |
|---|---|---|
| 10 | 168.0 | — |
| 20 | 134.9 | −19.7% |
| 30 | 110.9 | −34.0% |
| 40 | 93.2 | −44.5% |
| 60 | 69.7 | −58.5% |
| 80 | 55.2 | −67.1% |
| 100 | 45.5 | −72.9% |
Source: TRT Provider Guide model grid, calculated September 2, 2026, Vermeulen equation with albumin held at 4.3 g/dL. These are calculated scenarios, not measurements from real men, and this is a sensitivity calculation — it does not mean that changing a person's SHBG would produce this exact change in that person.
This is why two men with identical total testosterone can be in genuinely different situations, and it is the main reason guidelines suggest checking free testosterone when SHBG is likely to be unusual.
Things that tend to raise SHBG in men, per Labcorp's own summary: aging, an overactive thyroid, estrogens, HIV disease, anticonvulsant medication, and liver disease. Things that tend to lower it: obesity, diabetes, an underactive thyroid, steroid medication, androgen use, nephrotic syndrome, and acromegaly.
What does albumin change in the calculation?
Answer capsule: Albumin matters much less than SHBG. Across the normal albumin range of 3.5 to 5.0 g/dL, the calculated result moves by roughly 10 to 15%. But if you guess albumin instead of measuring it, that guess is silently baked into your answer.
Most calculators default to 4.3 g/dL, which is a reasonable population average. Ours does too, and it labels the assumption on screen when you use it.
Use your own albumin from the same blood draw if you have it. If you do not, know that you are reading an estimate built on top of an estimate.
pg/mL or ng/dL? The ten-times trap
Answer capsule: Free testosterone is reported in two common units and one is exactly ten times the other. 10 pg/mL equals 1 ng/dL. Comparing a result in one unit to a range written in the other is wrong by a factor of ten, and it is the most common mistake we see in published "normal ranges" online.
| From | To | Do this | Example |
|---|---|---|---|
| pg/mL | ng/dL | divide by 10 | 70 pg/mL = 7.0 ng/dL |
| ng/dL | pg/mL | multiply by 10 | 7.0 ng/dL = 70 pg/mL |
| pg/mL | pmol/L | divide by 0.2885 | 70 pg/mL ≈ 243 pmol/L |
| pmol/L | pg/mL | multiply by 0.2885 | 243 pmol/L ≈ 70 pg/mL |
Source: standard conversions using testosterone's molecular weight of 288.42 g/mol. The pmol/L factor is as stated by Jasuja et al., Andrology 2023;11:125–33.
We are not naming and shaming anyone, because the pattern is the point and it is easy to make. Check your own report's units before you compare your number to anything you read online, including this page.
When is free testosterone worth testing at all?
Answer capsule: Guidelines put it second, not first. The Endocrine Society recommends total testosterone as the initial test, with free testosterone reserved for men whose total is near the lower limit or who have a condition that shifts SHBG. The American Urological Association does not recommend free testosterone as the primary diagnostic method.
The Endocrine Society's position, in its 2018 clinical practice guideline, is that the first test should be total testosterone on a fasting morning sample, repeated to confirm. Free testosterone comes in when the total is borderline, or when something is likely to be moving SHBG around — thyroid disease, obesity, liver disease, certain medications.
The AUA is more blunt about the measured version. On direct free testosterone measurement, its Testosterone Deficiency Guideline says it is not reliable because of high CV — meaning the result varies too much from run to run.
Both positions are consistent with everything on this page. Free testosterone is a useful second look at a confusing picture. It is not a better first test, and no single free testosterone number decides anything on its own.
Why this page does not tell you whether you are low
We thought hard about whether to show a "below / inside / above the selected published interval" badge on the calculator, the way most tools do.
We decided not to, and the reason is Table 1. A verdict badge requires picking one reference range and treating it as the truth. There are at least seven of them for a man in his thirties, they disagree by a factor of 13.8, and no standards body has picked a winner.
So instead of one verdict, the calculator shows you what all seven published ranges would say about your number. Often they disagree with each other. That disagreement is not a bug in our tool. It is the most accurate thing we can tell you.
If you want to know what your result means for you specifically, that is a conversation with a clinician who can see your symptoms, your history, and a repeat morning test — not a number from any website, ours included.
Why this matters now
Free testosterone is being calculated more often than at any point in its history. It shows up on direct-to-consumer lab panels, in telehealth intake forms, and on dozens of free web calculators, and it is increasingly the number people bring to an appointment.
Meanwhile the measurement science is actively unsettled. The CDC ran a standardization study in 2026 and found methods varying by more than a factor of three. A major reference range study published in 2026 disagrees with a major reference range study published in 2023 by more than double at the lower limit. There is still no harmonized range.
That combination — a number being used more and more, while the ground under it is still moving — is why we thought this page needed to exist. Not to scare anyone off testing, but so that a person holding a lab report knows what kind of number they are holding.
Limitations of this analysis
We would rather state these plainly than have someone find them.
Our grid is a model, not a population. The 48 scenarios are calculations. No real man's blood is in them. They show how equations behave, not how men are distributed.
The reference ranges are not perfectly comparable. Age bands differ between sources — Labcorp's direct assay uses 20–29 while its calculated test uses 18–30. We matched them as closely as the published data allows and flagged it in Table 6.
Ranges differ for two reasons at once, and we cannot separate them. Method and reference population both vary between sources. Jasuja et al. studied healthy nonobese men. Labcorp cites internal studies. Ho et al. used 126 men in Edinburgh with normal semen analysis. Some of the spread in Table 1 is method; some is who got measured. We are not able to say how much is which.
One source is a conference abstract, not a paper. The CDC interlaboratory findings come from an ENDO 2026 abstract presented June 14, 2026. Abstracts are not peer reviewed to the standard of a full paper, and the numbers may shift when the full study publishes.
We did not implement the Zakharov model. We could not obtain its full parameter set from a primary source, so we describe its published performance rather than computing it. We publish no number we have not derived and checked ourselves.
Reference ranges change quietly. Lab test pages get revised without announcement. Every range on this page carries the date we checked it. If you are reading this long after September 2, 2026, verify the rows that matter to you.
We are not a clinical body. This page is an editorial research resource. It reports what published sources say and shows our arithmetic. It is not a guideline, not a diagnosis, and not medical advice.
How to cite this page
Citation information:
Page: TRT Provider Guide. “Free Testosterone Calculator: Five Published Methods and Reference Ranges Compared.” TRT Provider Guide Research. Last verified September 2, 2026. https://trtproviderguide.com/research/free-testosterone-calculator/
Dataset: TRT Provider Guide free testosterone model grid and reference range ledger, version 1.0, generated September 2, 2026. Forty-eight modeled scenarios across five implemented calculations, plus a published reference-range ledger and source CSV files.
The source line beneath each table and figure identifies the underlying source and verification date.
Frequently asked questions
What is a normal free testosterone level for a man?
There is no single agreed answer as of September 2, 2026. Across seven published reference ranges that apply to men in their thirties, the lower limit of normal runs from 8.7 to 120 pg/mL. Which one applies to you depends on how your lab produced the number. The Endocrine Society's guideline states that a harmonized reference range has not been established.
Is calculated free testosterone accurate?
It is a reasonable estimate with a known bias. In the largest published comparison we found, the Vermeulen calculation ran a median 1.19 times higher than directly measured free testosterone, with 95% of results between 0.90 and 1.47 times measured. It is useful. It is not the same as a measurement.
Why is my calculated free testosterone so much higher than the number on my lab report?
Most likely because your lab measured it with a direct immunoassay, which reads substantially lower. Labcorp's own page notes that the analog method returned results about a quarter as high as dialysis in the original comparison work. In 2026 the CDC measured immunoassay bias at −86.5% to −67.7% against its reference method. A calculated number and a direct-assay number are not on the same scale.
Do I need albumin to calculate free testosterone?
Two of the five equations do not use albumin at all. The three mass-action equations do. Most calculators substitute 4.3 g/dL if you do not have it, which is a fair population average but can be wrong for you if you have liver disease, kidney disease, or malnutrition. Use your own value from the same blood draw when you have it.
What is the difference between free and bioavailable testosterone?
Free testosterone is the slice held by nothing. Bioavailable testosterone is that slice plus the portion loosely held by albumin, which can also come off and be used. Bioavailable is always the larger number. In the ISSAM worked example, free testosterone was 5.02 ng/dL and bioavailable was 118 ng/dL.
Can free testosterone be low when total testosterone is normal?
Yes, and high SHBG is the usual reason. In our model grid, a man with a total testosterone of 500 ng/dL calculated to 168.0 pg/mL of free testosterone at an SHBG of 10, and to 45.5 pg/mL at an SHBG of 100 — a 72.9% difference with total testosterone unchanged. Those are calculated scenarios, not patient data.
Which free testosterone equation is best?
There is no settled answer. Vermeulen 1999 is the most widely used and is what most labs and calculators implement. In the Fiers comparison, the Ly–Handelsman equation had a median ratio closer to 1.00 against measured values but scattered more widely, while Vermeulen ran about 19% high with tighter scatter. Both remain in published use.
Does a free testosterone calculator diagnose low testosterone?
No. Guidelines treat total testosterone as the initial test, with free testosterone as a second look when total is borderline or SHBG is likely to be abnormal. A calculator output is an estimate from a formula. Diagnosis involves symptoms, repeat morning testing, and a clinician's assessment.
Sources
All links go to the original publisher. Verified September 2, 2026.
Guidelines and standards bodies
- Bhasin S, et al. Testosterone Therapy in Men With Hypogonadism: An Endocrine Society Clinical Practice Guideline. J Clin Endocrinol Metab 2018;103(5):1715–44. https://academic.oup.com/jcem/article/103/5/1715/4939465
- American Urological Association. Testosterone Deficiency Guideline. https://www.auanet.org/guidelines-and-quality/guidelines/testosterone-deficiency-guideline
- Zhou H, Ribera A, Sugahara O, Danilenko U, Vesper H. Assessment of Free Testosterone Measurement Methods: Findings from an Interlaboratory Comparison Study. CDC Clinical Standardization Programs. ENDO 2026 abstract SUN-144, presented June 14, 2026.
The equations
- Vermeulen A, Verdonck L, Kaufman JM. A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 1999;84(10):3666–72. PMID 10523012
- Södergård R, Bäckström T, Shanbhag V, Carstensen H. J Steroid Biochem 1982;16(6):801–10. PMID 7202083
- Nanjee MN, Wheeler MJ. Ann Clin Biochem 1985;22(4):387–90
- Ly LP, Handelsman DJ. Eur J Endocrinol 2005;152(3):471–8. PMID 15757865
- Ho CKM, Stoddart M, Walton M, Anderson RA, Beckett GJ. Calculated free testosterone in men: comparison of four equations and with free androgen index. Ann Clin Biochem 2006;43:389–97
- Zakharov MN, et al. A multi-step, dynamic allosteric model of testosterone's binding to sex hormone binding globulin. Mol Cell Endocrinol 2015;399:190–200. PMID 25240469
- ISSAM / Ghent University Hospital calculator and worked example. https://www.issam.ch/freetesto.htm
Reference ranges and method comparisons
- Fiers T, Wu F, Moghetti P, Vanderschueren D, Lapauw B, Kaufman JM. Reassessing Free-Testosterone Calculation by LC–MS/MS Direct Equilibrium Dialysis. J Clin Endocrinol Metab 2018;103(6):2167–74. PMID 29618085
- Jasuja R, et al. Reference intervals for free testosterone in adult men measured using a standardized equilibrium dialysis procedure. Andrology 2023;11(1):125–33. PMID 36251328
- Walravens J, et al. Age-Stratified Reference Ranges for Directly Measured Serum Free Testosterone in Community-Dwelling and Healthy Men. J Clin Endocrinol Metab 2026;111(3):e787–e793. PMID 40929374
- Winters SJ. Laboratory Assessment of Testicular Function. In: Endotext. Updated March 28, 2026. NCBI Bookshelf NBK279145, Table 4
- Labcorp test 144980, Testosterone, Free, Direct. https://www.labcorp.com/tests/144980/testosterone-free-direct
- Labcorp test 140226, Testosterone Free, Profile I. https://www.labcorp.com/tests/140226/testosterone-free-profile-i
- Michigan Medicine MLabs. Test Update 780, posted October 27, 2021. https://mlabs.umich.edu/2021/10/27/test-update-780-1
Related research from TRT Provider Guide
This page is an educational research resource published by TRT Provider Guide, an independent research and reference resource on testosterone replacement therapy. It is not medical advice, it does not diagnose any condition, and it does not recommend any treatment or provider. Free testosterone results should be interpreted by a licensed clinician alongside your symptoms, repeat morning testing, and your full medical history. No medical reviewer is claimed.