Why there is no "normal" number here
Almost every HOMA-IR calculator online will hand you a number and a verdict. The verdict is the part that isn't supported. There is no universally accepted cutoff for HOMA-IR. Thresholds in common use in the US fall somewhere between 2.0 and 3.0 — NHANES uses 2.5 — while cut points reported in Asian populations run lower. The score also isn't attached to any validated treatment algorithm: no guideline says "at this number, do this."
That doesn't make it useless. It makes it a trend instrument rather than a verdict machine. Your own HOMA-IR measured the same way, in the same lab, six months apart, is a far more informative number than your HOMA-IR compared to a threshold from a study population you may not resemble.
The caveat the original authors made and almost nobody repeats: the simple formula this page uses — HOMA1 — is described by Matthews and colleagues as a crude approximation of their actual model, reasonable near normal glucose and insulin values and less so away from them. It was derived from normal-weight adults under 35. The authors built a corrected computer model, HOMA2, recalibrated for modern insulin assays, and they recommend using it where possible. If your numbers are far from the normal range, treat the figure above as a rough orientation and the HOMA2 calculator as the better instrument.
Getting the inputs right
- Both samples must be fasting, and from the same draw. Pairing a fasting glucose with an insulin taken at a different visit produces a number that means nothing.
- Insulin assays are not standardised between labs. Two labs can return different insulin values for the same blood. This is the main reason to compare your score only against your own previous scores, from the same lab.
- Many standard panels don't include fasting insulin. You usually have to ask for it specifically — it isn't part of a routine lipid or metabolic panel.
- If you take insulin, HOMA-IR doesn't apply. The model assumes your own pancreas produced the insulin being measured.
What the number is made of
| Units | Formula |
|---|---|
| Glucose mg/dL | (fasting glucose × fasting insulin) ÷ 405 |
| Glucose mmol/L | (fasting glucose × fasting insulin) ÷ 22.5 |
Insulin is in µIU/mL in both (sometimes written mIU/L — the same thing). If your lab reports insulin in pmol/L, this page divides by 6.0 to convert. The model was scaled so a healthy young adult sits near 1.
What moves it is what moves insulin sensitivity generally: sleep, muscle, movement after meals and what the meal was made of. If you want the graded evidence on each of those rather than claims, the Evidence Library is the honest version.
Educational tool only — not a diagnostic tool, not medical advice, and not a substitute for the interpretation of your own clinician, who can see the rest of your results. Nothing you enter here is stored or sent anywhere.
Sources
- Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia. 1985;28(7):412–419. pubmed.ncbi.nlm.nih.gov/3899825
- Levy JC, Matthews DR, Hermans MP. Correct homeostasis model assessment (HOMA2) evaluation uses the computer program. Diabetes Care. 1998;21(12):2191–2192. pubmed.ncbi.nlm.nih.gov/9839117
- MDCalc. HOMA-IR — interpretation and cutoff caveats. mdcalc.com/calc/3120