The HOMA-IR test is not a needle in your arm but a short calculation: it takes your fasting blood sugar and your fasting insulin, multiplies them, and turns the result into one number that estimates how hard your body works to keep blood sugar steady. Doctors and researchers use it as a rough marker of insulin resistance, mostly in the liver. It is cheap and easy to repeat, which is why it appears on more and more lab panels. It is also widely misread. In this article you will learn how the score is built, what the numbers usually mean, why no single cutoff fits everyone, what moves your result up or down, and where the calculation stops being useful.
What the HOMA-IR test actually measures
HOMA-IR stands for Homeostatic Model Assessment of Insulin Resistance. It was designed in the 1980s to describe, with simple math, the balance between two values measured in the same fasting blood sample: the sugar circulating in your blood, and the insulin your pancreas releases to hold that sugar in place.
Insulin tells your liver to stop releasing stored sugar and tells muscle and fat cells to take sugar in. When those tissues respond less well, the pancreas compensates by making more insulin. For years, that extra insulin can keep blood sugar looking perfectly normal. The score is designed to catch exactly that situation: normal sugar, bought at the price of high insulin.
One detail matters. Because the calculation uses a fasting sample, it mostly reflects the overnight period, when the liver sets blood sugar. So it leans toward describing hepatic insulin resistance, meaning resistance in the liver, rather than in muscle. Muscle resistance shows up better after a meal. That is one reason two people with the same score can have different metabolic pictures. You can start with the sugar half of the equation by reading our guide to fasting glucose test results.
How the HOMA-IR formula works, in US and SI units
The math is short enough to do on a phone. The version you use depends on your lab’s units.
In US conventional units, where glucose is reported in milligrams per deciliter (mg/dL) and insulin in microinternational units per milliliter (uIU/mL), the formula is: fasting insulin multiplied by fasting glucose, divided by 405.
In SI units, where glucose is reported in millimoles per liter (mmol/L) and insulin in milliinternational units per liter (mIU/L), the formula is: fasting insulin multiplied by fasting glucose, divided by 22.5.
An example makes it concrete. With a fasting glucose of 92 mg/dL and a fasting insulin of 11 uIU/mL, you get 92 times 11, which is 1,012, divided by 405, giving roughly 2.5. That same glucose is about 5.1 mmol/L, and 5.1 times 11 divided by 22.5 gives the same answer. The two formulas are one equation in different units, so a calculator asking for mg/dL and one asking for mmol/L should agree. If they do not, check the unit you entered.
Two inputs, no more. That is both the strength and the weakness of the method: it needs only a standard fasting panel, but it can never be more reliable than those two measurements. You can see what the insulin half means on its own by reading our guide to fasting insulin blood test results.
Why there is no single normal HOMA-IR range
Search for a normal range and you will find figures from about 1.0 to well above 3.0, each presented with equal confidence. Three things explain the spread.
Insulin assays are not standardized
Unlike glucose, insulin measurement has never been fully harmonized between laboratories. Different immunoassays react differently to insulin fragments and to proinsulin, so one tube of blood can produce noticeably different insulin values in two labs. Since insulin is one of only two ingredients, that variation flows straight into the score. A 2.4 from one lab is not strictly comparable to a 2.4 from another.
Populations differ
Thresholds are usually set by looking at the distribution of scores in a healthy reference group, then drawing a line near the top of it. Change the group and the line moves. Studies in East Asian, South Asian, Latin American, European, and African-descent populations have produced clearly different cutoffs, partly because body composition and typical insulin secretion differ. Age matters too: adolescents are physiologically more insulin resistant during puberty, so adult numbers do not apply to them.
The purpose of the cutoff changes the cutoff
A threshold chosen to catch as many at-risk people as possible sits lower than one chosen to be certain. Research, screening, and clinical follow-up each need a different balance. This is why the index remains, in most guidelines, a research and monitoring tool rather than a diagnostic test.
Two versions of the model exist
The original 1985 equation has largely been replaced in research by an updated computer model, usually written HOMA2, which handles very high and very low values better and can run on C-peptide instead of insulin. The two are not interchangeable, and HOMA2 typically returns a lower figure than the classic formula for the same blood sample. Most lab reports and online calculators still use the original equation, so check which version produced your result before comparing it with a published threshold.
How to read your own HOMA-IR number
With those cautions in mind, the bands below reflect values commonly used in published research. They are orientation, not diagnosis, and they mean little without the rest of your results and your history.
| Approximate value | How it is usually read | Sensible next step |
|---|---|---|
| Below about 1.0 | Insulin sensitivity typical of lean, active reference groups | Nothing specific; keep the habits that got you there |
| About 1.0 to 2.0 | Common in the general adult population, unremarkable on its own | Read it with waist size, blood pressure, lipids and A1C |
| About 2.0 to 2.9 | A gray zone, above many research cutoffs but below others | Worth discussing with a clinician, ideally with a repeat sample |
| About 3.0 and above | Above most published thresholds, suggesting reduced insulin sensitivity | Medical review, with standard glucose testing for the fuller picture |
Look at what surrounds the score as well. A raised value alongside a wide waist, high triglycerides, low HDL cholesterol and blood pressure creeping upward describes a different level of concern than the same value sitting alone on an otherwise clean panel.
Read the trend rather than the decimal. A score drifting from 3.4 to 2.6 over a year, in the same lab under the same conditions, says more than a single value compared with a number you found online. A high score with a normal fasting glucose is also a different situation from a high score alongside rising blood sugar. You can follow three months of average blood sugar by reading our guide to hemoglobin A1C test results.
How HOMA-IR compares with other insulin resistance tests
No single blood test defines insulin resistance. Each method answers a slightly different question, and they are not interchangeable.
| Test | What it shows | Fasting | Main strength | Main limit |
|---|---|---|---|---|
| Fasting glucose | Blood sugar after an overnight fast | Yes | Standardized; used to diagnose prediabetes and diabetes | Stays normal for years while insulin quietly rises |
| Hemoglobin A1C | Average blood sugar over roughly three months | No | Convenient, stable, well standardized | Distorted by anemia and other red cell conditions |
| Fasting insulin | How much insulin the pancreas releases at rest | Yes | Can rise before blood sugar does | Assays vary between labs; no agreed normal range |
| HOMA-IR | Estimated insulin resistance, mostly hepatic, from those two values | Yes | Cheap and repeatable; useful for tracking change over time | No universal cutoff; inherits the insulin assay problem |
| Oral glucose tolerance test | How the body handles a measured sugar drink over two hours | Yes | Reveals after-meal problems that fasting tests miss | Takes hours; results vary from day to day |
| Euglycemic clamp | Glucose the body absorbs at a fixed insulin level, measured directly | Yes | The reference standard other methods are judged against | Research only: hours long, intravenous lines, specialized staff |
Notice where the index sits. It approximates something the clamp measures directly, using two ordinary lab values. That is useful for following one person over time. It is not a substitute for the tests that define prediabetes and diabetes. We look at that tension further in our piece on the gap between insulin resistance and A1C.
What raises HOMA-IR, and how the blood draw affects it
Before reading anything into a number, check how the sample was collected. The score is only valid on a true fasting sample, generally 8 to 12 hours with water allowed. Coffee, a bite of breakfast, or a very late dinner can move the insulin value enough to change the result. Hard exercise the night before, an acute illness, a poor night of sleep, or stress on the morning of the draw can all shift it. Repeat measurements in the same person often differ by a wide margin, which is why a single borderline score is rarely worth acting on.
Factors that tend to push the score up
- Visceral fat, meaning fat stored around the abdominal organs, which is more metabolically active than fat under the skin
- Long periods of sitting and low overall physical activity
- Short or broken sleep, and untreated sleep apnea
- Certain medications, including glucocorticoids, some antipsychotics, and some HIV therapies
- Pregnancy, where insulin resistance rises normally in the second and third trimesters
- Puberty, which temporarily reduces insulin sensitivity in healthy adolescents
- Hormonal conditions such as polycystic ovary syndrome, Cushing syndrome, and untreated thyroid disease
- Fat accumulation in the liver, which often travels with a raised score
Some of these are temporary and expected. Pregnancy and puberty are normal states, not diseases, and a score measured during either should not be compared with adult reference bands. Liver-related causes are worth checking with a clinician, and you can review the enzyme used most often for that in our guide to ALT liver enzyme levels. A second enzyme is covered in our explainer on gamma-glutamyl transferase results.
What tends to lower HOMA-IR
Insulin sensitivity responds to behavior faster than most lab values, often within weeks. None of the following is a prescription, and none replaces care from your own clinician, but these are the levers with the strongest evidence behind them.
- Regular physical activity. Aerobic exercise and resistance training both help, and a single session improves glucose uptake for a day or more, so consistency beats intensity.
- Modest weight loss. Around 5 to 7 percent of body weight is the target used in major diabetes prevention programs, and it is enough to shift metabolic markers meaningfully.
- Breaking up sitting time. Short walks every hour blunt the after-meal insulin response better than one long session at the end of a sedentary day.
- Sleep. Restoring adequate sleep and treating sleep apnea improve insulin sensitivity independently of weight.
- Diet quality. More fiber, whole grains, legumes, vegetables and unsaturated fats, with fewer sugar-sweetened drinks and refined starches. The pattern matters more than any single food.
- Alcohol moderation, which also lightens the load on the liver.
Blood fats usually move in the same direction, so a falling score often comes with lower triglycerides. Interpret the two together by checking our overview of triglyceride blood levels.
When to see a doctor, and what the number cannot tell you
A result here is a conversation starter, not a verdict. Book an appointment if your score sits clearly above the bands above, if it is rising across repeated tests, if your fasting glucose or A1C is also drifting upward, or if you have symptoms such as unusual thirst, frequent urination, blurred vision, or unexplained weight change. A family history of type 2 diabetes, a previous pregnancy with gestational diabetes, or polycystic ovary syndrome all lower the threshold for getting checked. The condition itself is described in our overview of type 2 diabetes symptoms and treatments.
Be equally clear about the limits. The calculation cannot diagnose diabetes or prediabetes; those rest on fasting glucose, A1C, or a glucose tolerance test. It cannot tell you whether your pancreas is losing the ability to make insulin, which other markers address, including the one covered in our guide to C-peptide test results. It says nothing about how your muscles handle sugar after a meal, and it does not predict who will develop diabetes at the level of one individual. In people already taking insulin, the formula breaks down entirely, because the insulin in the sample did not come from their pancreas. Sugar spilling into the urine is a separate signal, described on our page about glucose in a urine test.
Latest scientific advances
Research from the last three years has focused less on finding the perfect cutoff and more on what actually moves the number.
A 2025 systematic review pooled 87 studies covering more than 235,000 adults to estimate how common insulin resistance is worldwide. Roughly one in four adults met the criteria used in those studies. The more useful finding sits in the methods: the included studies used many different thresholds, and the overall estimate held up anyway. What this means for you is that the exact cutoff your lab prints matters less than where your number sits against your own earlier results. Standardized thresholds still do not exist.
A 2026 meta-analysis, a study that statistically combines many trials, reviewed 77 randomized trials of lifestyle programs in people with prediabetes, covering more than 22,000 participants. Losing at least 5 percent of body weight was linked to a clearly better chance of blood sugar returning to the normal range than smaller losses. Programs with supervised exercise sessions cut new diabetes cases more than advice alone, and younger participants improved their insulin measures more. What this means for you is that structure and supervision appear to matter, not just intention.
An analysis of the PREMIER trial, published in 2023, followed 685 adults for 18 months and separated the effects of diet quality, fitness and weight loss. Better diet quality and better fitness were each linked to lower insulin resistance, and part of that benefit remained after accounting for the weight people lost. What this means for you is that eating better and moving more seem to help even in a period when the scale is not moving.
A randomized trial published in 2025 compared aerobic exercise, resistance training, and both combined in 160 older adults with obesity who were also losing weight through diet. The combined group improved insulin sensitivity more than either type alone, while resistance training did the most to preserve muscle. What this means for you is that mixing cardio with strength work is a reasonable default. These are individual trials and pooled analyses in specific groups; they point in a direction rather than settling the question, and none of them changes how insulin resistance is formally diagnosed.
Glossary
| Term | Definition |
|---|---|
| HOMA-IR | Homeostatic Model Assessment of Insulin Resistance. A calculated score estimating insulin resistance from fasting glucose and fasting insulin in the same sample. |
| Insulin resistance | A state in which liver, muscle and fat cells respond less well to insulin, so the pancreas must release more of it to keep blood sugar in range. |
| Fasting insulin | The insulin circulating after 8 to 12 hours without food, reported in microinternational units per milliliter in the United States. |
| Hepatic | Relating to the liver. Hepatic insulin resistance means the liver keeps releasing sugar even when insulin signals it to stop. |
| Visceral fat | Fat stored deep in the abdomen around the organs, more closely tied to metabolic problems than fat just under the skin. |
| A1C | Hemoglobin A1C, a blood test reflecting average blood sugar over about three months. Used to diagnose and monitor diabetes. |
| Euglycemic clamp | A research procedure holding insulin and blood sugar at set levels to measure insulin sensitivity directly. Considered the reference standard. |
| Assay | The laboratory method used to measure a substance. Different insulin assays can give different values for the same sample. |
FAQ
Is there an official normal range for HOMA-IR?
No. There is no internationally agreed reference range, and laboratories and studies use different thresholds. Values around and below 1.0 are typical of lean, active reference groups, and most published cutoffs for reduced insulin sensitivity fall between 2.0 and 3.0. Because insulin assays are not standardized, the same blood can yield different scores in different labs. Treat any range you find as orientation, compare your result with your own earlier results from the same lab, and let a clinician read it alongside your glucose, A1C, blood pressure and history.
Can I calculate it myself from my lab report?
Yes, if your report includes fasting glucose and fasting insulin from the same draw. In US units, multiply insulin in microinternational units per milliliter by glucose in milligrams per deciliter, then divide by 405. In SI units, multiply insulin in milliinternational units per liter by glucose in millimoles per liter and divide by 22.5. Online calculators do exactly this, so the answers should match. The result only means something if the sample was truly fasting, and it is not valid if you take insulin injections.
Is HOMA-IR better than A1C?
They answer different questions, so neither replaces the other. A1C tells you what your blood sugar has actually been doing over roughly three months, and it is standardized and used for diagnosis. The index estimates how much insulin it took to achieve that blood sugar, which can change earlier. In practice it may flag a metabolic shift while A1C is still normal, but only A1C and glucose testing can establish prediabetes or diabetes. Most clinicians use glucose and A1C as the backbone and treat an insulin-based score as extra context.
Can I have insulin resistance with a normal A1C?
Yes, and it is common. Insulin resistance often develops years before blood sugar rises, because the pancreas compensates by making more insulin. During that period, fasting glucose and A1C can both look entirely normal while fasting insulin drifts upward. That is precisely the window this score was designed to describe. It is also why a normal A1C does not on its own rule out a metabolic problem, and why waist circumference, blood pressure and triglycerides are worth reviewing at the same visit.
Do I really have to fast before the test?
Yes. The formula assumes a fasting state, usually 8 to 12 hours with plain water allowed, and it becomes meaningless otherwise. Even a small amount of food raises insulin sharply and inflates the score. Keeping conditions consistent between tests helps too: same lab, similar time of morning, no unusually hard exercise the night before, no acute illness. Because repeat measurements in the same person vary noticeably, many clinicians confirm a borderline or unexpected result with a second sample.
How fast can the number change?
Faster than most blood markers. Insulin sensitivity responds to a single exercise session within a day, and studies of lifestyle programs typically see measurable change within 8 to 12 weeks. That responsiveness cuts both ways: a week of poor sleep, a course of steroids, or an illness can lift the number temporarily. Retesting after at least three months of consistent habits gives a fairer picture than checking every few weeks, and using the same laboratory removes one large source of noise.
Sources
- National Institute of Diabetes and Digestive and Kidney Diseases — Insulin Resistance and Prediabetes — nih.gov
- MedlinePlus, National Library of Medicine — Prediabetes — medlineplus.gov
- Cleveland Clinic — Insulin Resistance: What It Is, Causes, Symptoms and Treatment — my.clevelandclinic.org
- Centers for Disease Control and Prevention — Preventing Type 2 Diabetes — cdc.gov
- Ballena-Caicedo J, Zuzunaga-Montoya FE, Loayza-Castro JA, et al. — Global prevalence of insulin resistance in the adult population: a systematic review and meta-analysis — Frontiers in Endocrinology, 2025 — doi.org/10.3389/fendo.2025.1646258
- Rong J, Ho M, Zhou D, Chau PH — Identifying Promising Practices in Lifestyle Intervention Programs for the Prediabetes Population: A Meta-Analysis and Meta-Regression of Randomized Controlled Trials — Worldviews on Evidence-Based Nursing, 2026 — doi.org/10.1111/wvn.70152
- Gradinariu V, Ard J, van Dam RM — Effects of dietary quality, physical activity and weight loss on glucose homeostasis in persons with and without prediabetes in the PREMIER trial — Diabetes, Obesity and Metabolism, 2023 — doi.org/10.1111/dom.15160
- Colleluori G, Viola V, Bathina S, et al. — Effect of aerobic or resistance exercise, or both on insulin secretion, ciliary neurotrophic factor, and insulin-like growth factor-1 in dieting older adults with obesity — Clinical Nutrition, 2025 — doi.org/10.1016/j.clnu.2025.05.016
Further reading
- Track shorter-term blood sugar swings with our explainer on fructosamine test results.
- Assess the particle count behind your cholesterol number by consulting our guide to apolipoprotein B results.
- Understand the hormonal condition most often linked with a raised score by reading our overview of polycystic ovary syndrome symptoms and causes.
- Learn what your lab means by average blood sugar with our short definition of estimated average glucose.
- Check the protective cholesterol that often falls alongside insulin resistance by reading our guide to HDL cholesterol levels.
Understand your lab results with BloodSense
A HOMA-IR score means very little on its own and a great deal next to the rest of your panel. BloodSense reads your results in context, connecting fasting glucose, fasting insulin, A1C and triglycerides into a picture you can follow from one test to the next. It explains each value in plain language and shows you what changed. It does not diagnose anything and it does not replace your doctor, who remains the person who decides what happens next.
