HOMA-IR Calculator: Insulin Resistance from Fasting Insulin
HOMA-IR estimates insulin resistance from one fasting blood test: fasting glucose (mmol/L) × fasting insulin (µIU/mL) ÷ 22.5. The model was calibrated so that a normal reference adult scores 1. Higher numbers mean your body needs more insulin to hold fasting glucose steady. There is no official cut-off. The University of Oxford team behind HOMA says there are no defined normal or abnormal thresholds, and adult studies use cut-offs from about 1.8 to 3.8.
Units: glucose in mg/dL × 0.05551 = mmol/L. Insulin in µIU/mL, mU/L and mIU/L is the same number. Insulin in pmol/L ÷ 6.0 = µIU/mL, though some labs use 6.945.
HOMA-IR calculator
Enter your fasting glucose and fasting insulin from the same blood sample, in the units printed on your report. Everything is worked out on your device, and nothing is saved or sent anywhere.
HOMA-IR chart: where the cut-offs fall
No guideline sets a HOMA-IR cut-off, so the bands on this page are built around the cut-offs published in adult population studies. Those studies used different insulin tests, so no single line fits everyone.
Sources: Wallace 2004; Esteghamati 2010; Gayoso-Diz 2013; Geloneze 2009; Bonora 1998; Qu 2011; Labcorp and Mayo Clinic Laboratories insulin ranges. Full references at the end of this page.
| HOMA-IR | Band | What it usually means |
|---|---|---|
| Below 1.0 | Below 1 | Under the model’s reference point. Usually fits good insulin sensitivity, if glucose is normal. |
| 1.0 to 1.7 | Below the cut-offs | Under every published adult cut-off we found (the lowest is 1.78) |
| 1.8 to 2.5 | Borderline | Where cut-offs for spotting metabolic syndrome sit. Read it with your other results. |
| 2.6 to 3.7 | Raised | At or above many cut-offs used to define insulin resistance (2.6 to 3.46) |
| 3.8 or above | High | Above every adult cut-off we found (the highest is 3.8) |
Bands use your HOMA-IR rounded to 1 decimal place. They give context, not a diagnosis.
How HOMA-IR is calculated
The formula comes from the homeostasis model assessment (HOMA), published by Matthews and colleagues at Oxford in 1985. The simple version, now called HOMA1, is:
HOMA-IR = fasting glucose (mmol/L) × fasting insulin (µIU/mL) ÷ 22.5
In US units the same formula reads glucose (mg/dL) × insulin ÷ 405, because 22.5 × 18 = 405. The calculator converts mg/dL with the factor 0.05551 (Mayo Clinic Laboratories), so its answer can differ from the ÷ 405 shortcut in the second decimal place.
Worked example: a fasting glucose of 5.0 mmol/L (90 mg/dL) and a fasting insulin of 10 µIU/mL give 5.0 × 10 ÷ 22.5 = 2.22, which rounds to 2.2.
Converting insulin: pmol/L and µIU/mL
µIU/mL, mU/L and mIU/L give the same number (Labcorp). Converting to pmol/L is less tidy. The CDC’s NHANES survey and the NIH’s Diabetes in America use 6.0 pmol/L per µIU/mL, and a 2019 review calls 6.00 the correct factor. Labs still differ: Mayo Clinic Laboratories lists 6.95, and Quest Diagnostics 7.175. An insulin of 60 pmol/L is 10 µIU/mL at 6.0 but 8.6 at 6.945, so the example above gives 2.2 or 1.9. The calculator uses 6.0 unless you choose another factor, and shows the difference.
HOMA2, the updated model
HOMA1 is a straight-line shortcut. The Oxford team later released HOMA2, a computer model recalibrated for current insulin tests that can also use C-peptide. Its calculator page says the simple equations “are not appropriate for use with currently available insulin assays”, and the HOMA2 calculator is meant for health professionals. HOMA2 gives lower numbers: in one Brazilian study, the insulin resistance cut-off was 2.7 with HOMA1 and 1.8 with HOMA2. This calculator gives HOMA1, so check which version a report or study uses before comparing.
What HOMA-IR actually measures
Insulin resistance means your body doesn’t respond to insulin the way it should (NIDDK). To keep glucose normal, the pancreas makes more insulin. As long as it can keep up, blood sugar stays in a healthy range. Eventually it can’t, and glucose rises (Cleveland Clinic). Our insulin resistance guide covers causes and signs.
A good detective judges not only the outcome but how hard someone had to work for it. Fasting glucose is the outcome and fasting insulin is the effort, so a normal glucose held with a lot of insulin gives a high HOMA-IR. In the 1985 study, HOMA’s estimate correlated closely with the euglycemic clamp, a detailed research test (Rs = 0.88), but its precision was modest, with a coefficient of variation of 31%.
What your HOMA-IR result means
Find your band. Each one sets out what the number usually suggests, the next lead your doctor may follow, and questions worth asking.
HOMA-IR below 1.0
Below the model’s reference point. Your fasting insulin is low for your glucose, which usually fits good insulin sensitivity.
A low score with a high glucose means something different. Insulin levels fall in type 1 diabetes and later in type 2 (Mayo Clinic Laboratories), so a low HOMA-IR can reflect too little insulin.
Usually nothing more for HOMA-IR. If you’re being checked for diabetes, guidelines use HbA1c or fasting glucose (ADA 2026, NICE PH38).
Ask: Are my glucose and HbA1c normal too? Is there any reason to repeat this test?
HOMA-IR 1.0 to 1.7
Above the reference point of 1, but below the lowest adult cut-off we found (1.78, from a national survey in Iran). On its own, this doesn’t point to insulin resistance.
The cut-offs come from studies using different insulin tests, so a result near 1.7 at one lab could read higher at another. See what can throw it off.
If you have risk factors such as a large waist, a family history of diabetes or PCOS (NIDDK), the usual screening tests are HbA1c and fasting glucose.
Ask: Do my risk factors mean I should be screened for prediabetes, and how often?
HOMA-IR 1.8 to 2.5
This is where cut-offs for spotting metabolic syndrome sit: 1.78 in Iran, 2.3 in Brazil, and in Spain 1.85 to 2.05 for men and 2.07 to 2.47 for women, depending on age. The Spanish values sit at the 70th to 75th percentile of the adults studied.
Metabolic syndrome, also called insulin resistance syndrome, means three or more of a large waist, high blood pressure, high blood sugar, high triglycerides and low HDL (NHLBI).
Your waist, blood pressure, cholesterol panel and HbA1c or fasting glucose. When HOMA is used for one person, its authors suggest three insulin samples, because one result varies.
Ask: Do I have other features of metabolic syndrome? Would a repeat test change the plan?
HOMA-IR 2.6 to 3.7
At or above many cut-offs used to define insulin resistance: 2.6 (described as a popular clinical cut-off), 2.7 (top 10% of healthy Brazilian adults), 2.77 (top 20% of non-obese Italian adults without metabolic disorders) and 3.46 (top 10% of a Spanish adult population).
In Italy’s Bruneck Study, insulin resistance defined this way was found in 84% of people with high triglycerides, 88% with low HDL, 66% with impaired glucose tolerance and 63% with high uric acid.
HbA1c or fasting glucose to check for prediabetes, a lipid panel, and liver tests if fatty liver is suspected. NIDDK lists fatty liver among problems linked with insulin resistance.
Ask: Do my other results point the same way? Am I at higher risk of type 2 diabetes, and how often should I be screened?
HOMA-IR 3.8 or above
Higher than every adult cut-off we found. The highest, 3.8, comes from a study of Mexican American adults in Texas and from a Spanish study.
A sample that wasn’t fasting, insulin treatment, or a unit mix-up such as pmol/L entered as µIU/mL.
HbA1c, fasting glucose or an oral glucose tolerance test, and a look at linked conditions such as fatty liver and PCOS. In the Diabetes Prevention Program, losing 5% to 7% of body weight lowered the chance of type 2 diabetes (NIDDK).
Ask: Could anything have pushed this result up? Which tests do I need next? What changes would help most in my case?
HOMA-IR vs HbA1c, fasting glucose and TG/HDL ratio
| Test | What it shows | Strength | Limit |
|---|---|---|---|
| HOMA-IR | How much insulin it takes to hold fasting glucose steady | Can flag insulin resistance while glucose is still normal | Insulin tests aren’t standardized, and there’s no official cut-off |
| HbA1c | Average blood glucose over about 3 months (NIDDK) | No fasting needed. Used to diagnose diabetes and prediabetes. | Can stay normal while extra insulin keeps glucose in range |
| Fasting glucose | Blood sugar after at least 8 hours without food | ADA prediabetes 100 to 125 mg/dL. NICE high risk 5.5 to 6.9 mmol/L. | The same blind spot as HbA1c early on |
| TG/HDL ratio | Triglycerides divided by HDL, from a standard cholesterol test | Cheap and widely available | Depends on the units used, and was studied in overweight adults |
In a 2003 study of 258 overweight adults, a triglyceride to HDL ratio of 3.0 or more, with both measured in mg/dL, helped pick out those who were insulin resistant. In mmol/L the same ratio works out at about 1.3, because the two fats use different conversion factors. The researchers suggested the triglyceride or TG/HDL cut-offs as the most practical approach, “in the absence of a standardized insulin assay”.
What can throw off a HOMA-IR result
- Not fasting. HOMA is a fasting model: the Oxford team says only values seen in a fasting person should be used. Mayo Clinic Laboratories requires 8 hours of fasting for its insulin test.
- The insulin test itself. Insulin tests aren’t standardized. When 10 commercial methods were compared with a reference method, only 4 had 95% of results within 32% of it (Miller 2009). A 2025 comparison of 12 tests still found significant variability.
- The conversion factor. For insulin in pmol/L, dividing by 6.0 instead of 6.945 gives a HOMA-IR about 16% higher.
- Insulin treatment. Some insulin tests don’t detect insulin analogues such as lispro, aspart and glargine, and antibodies to insulin can interfere (Mayo Clinic Laboratories). The HOMA authors say its use in people on insulin needs further validation.
- Medicines. Gliclazide increases how much insulin your body makes, and semaglutide, a GLP-1 medicine, helps it make more when needed (NHS). Long-term steroids (glucocorticoids), some antipsychotics and some HIV medicines raise the risk of insulin resistance itself, as does Cushing’s syndrome, a condition of excess cortisol (NIDDK).
- Biotin and hemolysis. Labcorp asks you to stop high-dose biotin at least 72 hours before an insulin test, and Mayo Clinic Laboratories asks for 12 hours without biotin supplements. Hemolysis, when red cells break open in the tube, interferes with insulin tests.
- One sample is a snapshot. When HOMA is used for an individual, its authors suggest taking three insulin samples.
- Mismatched samples. Glucose and insulin should come from the same fasting draw. Gloucestershire Hospitals NHS labs require a glucose sample with every insulin request.
Pregnancy, PCOS and other special groups
- Pregnancy: hormonal changes of pregnancy make the body less able to use insulin (NIDDK), so adult cut-offs don’t apply. Gestational diabetes is checked between 24 and 28 weeks with a glucose challenge test or an oral glucose tolerance test.
- PCOS: the 2023 international PCOS guideline says insulin resistance is part of PCOS, but “clinically available insulin assays are of limited clinical relevance and are not recommended in routine care.” It recommends a 75 g oral glucose tolerance test to check blood sugar.
- Sex and age: in the Spanish study, the best cut-off for men was 1.85 or 2.05, depending on the definition of metabolic syndrome. For women it rose from 2.07 at age 50 to 2.47 at age 70, and HOMA-IR was less accurate in older women.
- Ethnicity: the Spanish authors note that cut-offs differ by ethnicity. The Texas study of Mexican American adults found 3.8.
- Children and teenagers: the bands here come from adult studies, so ask a pediatric clinician how to read a child’s result.
US vs UK: units and how doctors assess insulin resistance
| Result | US report | UK report |
|---|---|---|
| Fasting glucose | mg/dL, e.g. 90 | mmol/L, e.g. 5.0 |
| Fasting insulin | µIU/mL (Labcorp and Mayo range: 2.6 to 24.9) | pmol/L at some labs (e.g. Gloucestershire), mIU/L at others (e.g. South Tees: 2.0 to 25.0) |
| HOMA-IR | The same number in both countries once the units are converted | |
In the US, labs such as Labcorp and Mayo Clinic Laboratories offer fasting insulin, but the ADA diagnoses prediabetes and diabetes with HbA1c, fasting glucose or an oral glucose tolerance test. The NIDDK says health care professionals may not test for insulin resistance, and that the test for it is used mainly in research.
In the UK, NHS labs mainly use insulin to investigate unexplained low blood sugar, and University Hospitals Birmingham doesn’t usually run it if glucose is above 3.1 mmol/L. NICE (PH38) asks GPs to start with a risk assessment, then a fasting glucose or HbA1c: 5.5 to 6.9 mmol/L or 42 to 47 mmol/mol (6.0 to 6.4%) counts as high risk.
The thresholds differ, too. The ADA’s prediabetes range starts at 100 mg/dL (5.6 mmol/L), while NICE’s high-risk range starts at 5.5 mmol/L (99 mg/dL). See our blood sugar levels guide.
Reading HOMA-IR on your lab report or app
- Names: usually “insulin” or “insulin, serum”, taken fasting. Some reports add a calculated HOMA-IR line.
- Units: a pmol/L result treated as µIU/mL makes HOMA-IR about six times too high. Check the unit next to the number.
- Flags: an insulin inside the lab range can still give a high HOMA-IR. With a glucose of 5.0 mmol/L (90 mg/dL), any insulin from 2.6 to 24.9 µIU/mL gives a HOMA-IR from 0.6 to 5.5.
- Version: check whether a printed score is HOMA1 or HOMA2 before comparing it with a cut-off.
- Apps and portals: in the US, results often appear in your patient portal before your clinician has commented. In the UK, results in the NHS App may come with a short note from your GP practice.
When to get help urgently
- Call 999 (UK) or 911 (US), or go to the emergency department, if you have symptoms of diabetic ketoacidosis: feeling very thirsty, peeing more often, stomach pain or being sick, breathing more deeply than usual, fruity-smelling breath, or feeling very sleepy or confused.
- A fasting glucose of 7.0 mmol/L (126 mg/dL) or more, especially with tiredness, thirst, peeing more or weight loss without trying: see your doctor promptly.
- Symptoms of low blood sugar, such as shaking or sweating, when you don’t have diabetes: see a doctor. Call 999 or 911 if someone with very low blood sugar isn’t responding normally.
HOMA-IR questions people ask
What is a normal HOMA-IR?
There isn’t an official one. The Oxford team behind HOMA says there are no defined normal or abnormal thresholds, because results depend on the lab’s tests. The model’s reference point is 1, and adult studies place cut-offs between about 1.8 and 3.8.
What is a good HOMA-IR score?
Lower generally means more sensitive to insulin, and under 1.8 is below every adult cut-off we found. No guideline sets a target, and a very low score with a high glucose can mean too little insulin.
How do I convert fasting insulin from pmol/L to µIU/mL?
Divide by 6.0, the factor the CDC and NIH use. Some labs use 6.945 or 7.175 instead. So 72 pmol/L is 12.0 µIU/mL at 6.0, or 10.4 at 6.945.
Can I calculate HOMA-IR with glucose in mg/dL?
Yes. Multiply mg/dL by 0.05551 to get mmol/L, or use glucose (mg/dL) × insulin ÷ 405. The calculator accepts both units.
Is HOMA-IR better than HbA1c?
They answer different questions. HbA1c shows your average glucose over about 3 months and is used to diagnose diabetes. HOMA-IR estimates how hard your body works to keep fasting glucose normal, so it can be high while HbA1c is still normal.
Why won’t my NHS GP test my insulin?
NHS labs mainly use insulin to investigate low blood sugar, and NICE’s approach to diabetes risk uses a risk assessment, fasting glucose and HbA1c. The 2023 international PCOS guideline also says insulin tests aren’t recommended in routine care.
Can I use HOMA-IR if I take insulin or diabetes medicines?
Only with caution. Some insulin tests miss injected insulin analogues, the HOMA authors say its use in people on insulin needs further validation, and gliclazide raises insulin output. Ask your diabetes team how to read it.
How we wrote this page
Written and medically reviewed by Dr. Mustafa Sakr, MBBCh. Last reviewed: October 2026. The formula comes from Matthews 1985 and the University of Oxford’s HOMA resources, the unit factors from US lab tables, the CDC and the NIH, and the bands from published adult population studies, all listed below. We update this page when guidance changes.
How the calculator works: it converts glucose to mmol/L (mg/dL × 0.05551) and insulin to µIU/mL (pmol/L ÷ 6.0, or the factor you choose), multiplies them and divides by 22.5. The result is rounded to 1 decimal place, with halves rounded up, and the band is based on that rounded number. It flags glucose in the diabetes range, unusual insulin values and likely unit mix-ups. Everything runs on your device. Nothing is stored or sent, and the calculator can’t take account of your symptoms, medicines or history. More calculators are on our tools page.
References
- 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. Oxford University Research Archive
- Wallace TM, Levy JC, Matthews DR. Use and abuse of HOMA modeling. Diabetes Care 2004;27(6):1487–1495. Full text (PDF)
- University of Oxford, Diabetes Trials Unit. HOMA2 Calculator and HOMA frequently asked questions. Calculator · FAQ
- Gayoso-Diz P, et al. Insulin resistance (HOMA-IR) cut-off values and the metabolic syndrome in a general adult population: effect of gender and age. EPIRCE cross-sectional study. BMC Endocr Disord 2013;13:47. doi:10.1186/1472-6823-13-47
- Other HOMA-IR cut-off studies: Esteghamati A, et al. Nutr Metab 2010;7:26 (Iran) DOAJ; Geloneze B, et al. Arq Bras Endocrinol Metab 2009;53(2):281–287 (Brazil) SciELO; Bonora E, et al. Diabetes 1998;47:1643–1649 (Italy) University of Parma; Qu HQ, et al. PLoS One 2011;6(6):e21041 (Mexican American adults) PLOS ONE.
- Knopp JL, Holder-Pearson L, Chase JG. Insulin units and conversion factors: a story of truth, boots, and faster half-truths. J Diabetes Sci Technol 2019;13(3):597–600. doi:10.1177/1932296818805074
- Centers for Disease Control and Prevention. NHANES 2017–2018 Insulin laboratory data documentation; and NIDDK, Diabetes in America (3rd edition) conversions. CDC · NIDDK (PDF)
- Miller WG, et al. Toward standardization of insulin immunoassays. Clin Chem 2009;55(5):1011–1018 abstract; and Rohlfing C, et al. The current status of serum insulin measurements and the need for standardization. Clin Chem Lab Med 2025 doi:10.1515/cclm-2025-0552
- American Diabetes Association. 2. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes 2026. Diabetes Care 2026;49(Suppl 1):S27–S49, summarised at diabetes.org; and NICE. Type 2 diabetes: prevention in people at high risk (PH38), 2012, updated 2017. nice.org.uk (PDF)
- Teede HJ, et al. Recommendations from the 2023 International Evidence-based Guideline for the Assessment and Management of Polycystic Ovary Syndrome. Hum Reprod 2023;38(9):1655–1679. Full text (PDF)
- McLaughlin T, et al. Use of metabolic markers to identify overweight individuals who are insulin resistant. Ann Intern Med 2003;139(10):802–809. Abstract (Stanford Health Care)
- US health pages: NIDDK insulin resistance and prediabetes, A1C test and gestational diabetes; NHLBI metabolic syndrome; Cleveland Clinic insulin resistance.
- US laboratory pages: Labcorp insulin (004333) and SI conversion table; Mayo Clinic Laboratories insulin, serum and SI conversion table; Quest Diagnostics SI units (PDF).
- UK pages: NHS labs at Gloucestershire Hospitals, University Hospitals Birmingham and South Tees; NHS.uk diabetic ketoacidosis, type 2 diabetes symptoms, low blood sugar, gliclazide and semaglutide.
This page is for education only and isn’t medical advice, a diagnosis or a substitute for your own doctor, who knows your history. Insulin tests and reference ranges vary between laboratories. Always interpret results with a qualified clinician. See our medical disclaimer.
