Insulin Blood Test: Normal Levels, High Insulin & HOMA-IR

Insulin is the hormone that moves sugar (glucose) out of your bloodstream and into your cells. Measuring it — usually as fasting insulin — asks how hard your pancreas is working to keep your glucose normal. A high insulin level most often points to insulin resistance. This guide covers normal insulin levels, what high and low results mean, how to calculate HOMA-IR correctly in U.S. units, and something lab marketing rarely admits: there is no agreed cutoff for fasting insulin, and this is not how diabetes gets diagnosed — that belongs to glucose and A1C.

Key takeaways

• Insulin is made by the beta cells of your pancreas and lowers blood glucose by letting cells take sugar in.¹
• There is no universal normal range for fasting insulin — values vary by assay, lab, and population. Use the range on your report.²³
• HOMA-IR in U.S. units: (fasting insulin µIU/mL × fasting glucose mg/dL) ÷ 405. The ÷ 22.5 version is only for glucose in mmol/L.²³
• No consensus HOMA-IR cutoff exists either: published thresholds run from 2.0 to 5.9.²
• Insulin testing is not routine in the U.S. The NIDDK is blunt — the test for insulin resistance "is primarily used only for research studies."⁴
• High fasting insulin usually means insulin resistance: obesity, metabolic syndrome, prediabetes, PCOS, fatty liver. Rarely an insulinoma.⁵⁶
• Low insulin with high glucose suggests the pancreas can't make enough, as in type 1 diabetes; C-peptide is the better test there.¹⁷ Insulin does not diagnose diabetes — glucose and A1C do.⁴

What is insulin?

Insulin is a hormone made by the beta cells of the pancreas. After you eat, glucose rises; the pancreas releases insulin, which acts like a key opening your muscle, liver, and fat cells so glucose can enter. Blood sugar falls.¹

When cells stop responding to that signal, you have insulin resistance: "cells in your muscles, fat, and liver don't respond well to insulin."⁴ The pancreas compensates by making more insulin — hyperinsulinemia.⁸ While it keeps up, glucose stays normal at the cost of a high insulin level — exactly what a fasting insulin can reveal.⁹² This is common: NHANES analyses put insulin resistance at about 22% of U.S. adults over 20, and 40% of adults aged 18 to 44 by HOMA-IR;⁵ about 97.6 million U.S. adults had prediabetes in 2021.⁴

Insulin and C-peptide. Making insulin also releases an equal amount of C-peptide, which "stays in your blood longer" and "is also unaffected by outside sources of insulin such as the medicines involved in diabetes treatment" — so it measures your own production even in someone taking insulin.⁷

Why is insulin measured — and why usually isn't it?

Here is the honest version, which differs sharply from what direct-to-consumer lab panels and "metabolic health" marketing suggest. Fasting insulin is not a routine U.S. test. The NIDDK states plainly: "Health care professionals may not test for insulin resistance. The test for insulin resistance is primarily used only for research studies."⁴ Cleveland Clinic agrees — "insulin resistance is difficult to diagnose because there isn't routine testing for it" — and the tests providers order instead are a glucose test, an A1C, and a lipid panel.⁸ StatPearls confirms such measures "have not been integrated into clinical guidelines."⁵

So when is it useful? The strongest indication is one people rarely hear: unexplained hypoglycemia, which MedlinePlus lists as the primary reason for the test.¹ High insulin while blood sugar is low is what raises suspicion of an insulinoma.⁶¹⁰ Insulin may also support a PCOS workup¹¹ or assess pancreatic reserve alongside C-peptide.⁷¹² None of that makes it a screening test.

Fasting insulin

Fasting insulin means insulin measured after 8 to 12 hours without food.¹ Fasting matters because insulin is dynamic: it spikes after every meal, so a random level is close to meaningless. Fasting standardizes it to a baseline — and it's the number HOMA-IR needs, which is why it's nearly always drawn with a fasting glucose. One practical detail: biotin supplements must be stopped at least a day beforehand, because they interfere with the assay.¹

The limitation is built in: a fasting index captures one moment. That's partly why the hyperinsulinemic-euglycemic clamp — "the gold standard method for the measurement of peripheral (muscle) IR" — remains a research procedure.²⁵

Normal insulin levels

Here honesty beats a tidy table. There is no universally agreed normal range for fasting insulin.

The reason is technical but decisive: unlike glucose, insulin assays are not standardized between laboratories. Gastaldelli's review quantifies it — "there is large interlaboratory variation in the measurement of insulin concentrations (25%)," versus roughly 5–10% for glucose or triglycerides, which "are standardized among laboratories."² Two labs can measure the same blood and disagree by a quarter.

Reference intervals are also population-specific. A 2026 BMJ Open study in healthy Nepalese adults derived a fasting insulin range of 2.63–14.56 µIU/mL and a HOMA-IR range of 0.56–3.50, arguing Western-derived intervals misclassify South Asian populations.³ That study exists precisely because one universal number doesn't.

What this means for you: if a website, supplement seller, or lab panel says your "optimal" fasting insulin should be under some specific number, ask where that number came from. Often, nowhere in particular.

HOMA-IR: the formula, and the unit trap

HOMA-IR (Homeostatic Model Assessment of Insulin Resistance), first described by Matthews and colleagues in 1985, combines fasting glucose and fasting insulin into a single estimate of insulin resistance — "a widely utilized measure of insulin resistance in clinical research."¹³⁵

The formula in U.S. units

This is the most important technical point here. The formula changes with the unit of glucose:

U.S. units (glucose in mg/dL): HOMA-IR = (fasting insulin µIU/mL × fasting glucose mg/dL) ÷ 405

SI units (glucose in mmol/L): HOMA-IR = (fasting insulin µIU/mL × fasting glucose mmol/L) ÷ 22.5

Gastaldelli's review gives both: (I0 mU/mL × G0 mg/dL)/405 and (I0 mU/L × G0 mmol/L)/22.5.² Independent sources reproduce the U.S. form exactly: "HOMA1-IR = (fasting insulin (µIU/mL) × fasting glucose (mg/dL))/405."³¹⁴

The two constants aren't rival conventions — they're the same equation. Since mg/dL ÷ 18 = mmol/L, and 405 ÷ 18 = 22.5, the math is identical once units line up.

The trap: plug a U.S. report in mg/dL into the European ÷ 22.5 formula and your result is wrong by a factor of about 18 — turning a normal value into an alarming one. Calculators that never ask which glucose unit you're using are the usual culprit.

A worked example

Say your report shows fasting insulin 12 µIU/mL and fasting glucose 95 mg/dL:

HOMA-IR = (12 × 95) ÷ 405 = 1,140 ÷ 405 = 2.8

Check it in SI units: 95 mg/dL ÷ 18 = 5.28 mmol/L, and (12 × 5.28) ÷ 22.5 = 2.8. Same answer — as it must be. Now the error: using ÷ 22.5 with mg/dL gives (12 × 95) ÷ 22.5 = 50.7, which is physiologically nonsensical. If your result lands in double digits, you almost certainly used the wrong constant.

What the number means — and its limits

Higher HOMA-IR means more probable insulin resistance. But there is no agreed threshold. Gastaldelli lists cutoffs including > 4.65 in a general population, > 3.60 when BMI exceeds 27.5, > 2.0 in fatty liver disease, and > 5.9 elsewhere.² Same index, same units — and someone scoring 3.0 is "resistant" under one threshold and fine under another.

Two limits worth holding onto. It inherits the assay problem: HOMA-IR is built from insulin, so that 25% interlaboratory variation flows straight into your result.² And it is a research tool — its value lies in tracking change over time at the same lab, not in diagnosing an individual.⁵

High insulin levels

MedlinePlus frames the core scenario: "If your insulin level is high and your blood glucose is normal or a little above normal for you, you may have insulin resistance."¹ The usual company it keeps:⁵

• Overweight and obesity, and metabolic syndrome (large waist, high blood pressure, high triglycerides, low HDL);
• Prediabetes — insulin resistance typically precedes type 2 diabetes by years;
• PCOS — see below;
• Fatty liver disease (MASLD), often with a raised ALT or GGT;
• cardiovascular disease, and rarely an insulinoma.⁶

A visible clue sometimes accompanies it: acanthosis nigricans, "darkened skin in your armpit or back and sides of your neck."⁸

"High insulin but normal glucose" is the signature of early insulin resistance: the pancreas floods the system with extra insulin, and glucose stays normal because of that effort.⁹² A high insulin can therefore be an early signal, appearing before glucose or A1C move — the legitimate kernel behind the fasting-insulin enthusiasm online. The catch: without a validated cutoff or a standardized assay, that signal is hard to act on.⁵⁴ A high insulin isn't a diagnosis — it's a clue, read against your weight, waist, glucose, A1C, and lipids.

PCOS and insulin

Polycystic ovary syndrome is where insulin resistance is most clinically central — and a genuine reason a clinician may look at insulin.¹¹ The NICHD describes the self-reinforcing loop: "the body produces more and more insulin to get glucose into the cells. To balance out the high levels of insulin, the body makes more androgens, which contribute to symptoms of PCOS."¹⁵ That loop explains why PCOS management so often targets insulin sensitivity rather than only the ovaries, and why PCOS sits on the NIDDK's list of risk factors.⁴

Insulinoma: rare, but the real indication

An insulinoma is a pancreatic tumor secreting insulin autonomously, causing hypoglycemia — high insulin while glucose is low. MedlinePlus notes insulinomas "are uncommon and usually aren't cancer"; StatPearls gives an incidence of "approximately 1 to 32 cases per million persons per year."¹⁶¹⁶

Suspicion starts with Whipple's triad: "symptoms of hypoglycemia, documented low plasma glucose levels, and resolution of symptoms following glucose administration." Diagnosis rests on a supervised 72-hour fast, where the findings are plasma glucose < 55 mg/dL with insulin ≥ 3 µU/mL, C-peptide ≥ 0.6 ng/mL, and a negative sulfonylurea screen.⁶¹⁰ Note what that is: a hospital protocol, interpreted as a set — not a fasting insulin drawn at a walk-in lab.

Low insulin levels

At fasting, a low insulin is often simply normal — the pancreas is idling. What matters is the glucose beside it. MedlinePlus: "If your insulin level is low and your blood glucose is high for you, it may mean that your pancreas can't make enough insulin," which can point to type 1 diabetes or pancreatitis.¹ There, clinicians lean on glucose, A1C, and C-peptide, which gauges remaining pancreatic output.⁷¹²

Insulin vs. glucose vs. A1C

The distinction that prevents most misreadings: fasting glucose and A1C diagnose and monitor diabetes — NIDDK prediabetes thresholds are A1C 5.7% to 6.4%, fasting glucose 100 to 125 mg/dL, OGTT 140 to 199 mg/dL.⁴ Insulin and HOMA-IR describe the mechanism (resistance versus deficiency), mainly in research.⁵² C-peptide measures your pancreas's own output.⁷

What can affect your insulin level

Time since your last meal, weight and waist, physical activity, stress and acute illness, pregnancy, biotin supplements,¹ medications such as corticosteroids, and above all which assay your lab runs — which is what makes comparing results across two labs unreliable.²

Recent research

According to recent PubMed publications:

• The assay problem is the bottleneck. Reviews stress that the euglycemic clamp remains the gold standard, while fasting indices carry 25% interlaboratory variation and a scatter of population-specific cutoffs rather than one threshold² — which is why work on population-specific reference intervals continues.³ A meta-analysis nonetheless links elevated HOMA-IR to adverse outcomes in people with metabolic risk factors, supporting its epidemiological value.¹⁷
• What reverses insulin resistance is well established. In the Diabetes Prevention Program, a lifestyle program targeting 7% weight loss and 150 minutes of activity per week cut type 2 diabetes incidence by 58%, and metformin by 31% — lifestyle beating the drug.¹⁸ NIDDK mirrors it: losing 5% to 7% of starting weight reduces the chance of developing type 2 diabetes.⁴ Trials continue testing insulin-sensitizing approaches in PCOS.¹⁹

These findings concern understanding and monitoring; they do not authorize self-medication and do not replace your physician's advice.

Get your insulin interpreted by AI DiagMe

An insulin level is never read alone: its meaning depends on your glucose, your A1C, your weight and waist, and your HOMA-IR — and the value itself shifts by 25% depending on which lab ran it.

👉 AI DiagMe interprets your lab results — blood, urine, or stool — in plain language, taking your whole profile into account. An informational service that does not provide a diagnosis and complements, never replaces, your physician.

Frequently asked questions

Bottom line

A fasting insulin mostly reflects how hard your pancreas is working — which is why a high insulin usually means insulin resistance (weight, metabolic syndrome, prediabetes, PCOS, fatty liver) and only rarely an insulinoma. Two things set this marker apart. First, there is no consensus normal range for fasting insulin and no agreed HOMA-IR cutoff: assays vary by about 25% between labs, and thresholds run from 2.0 to 5.9. Second, U.S. authorities are candid that this isn't a routine test. If you calculate HOMA-IR from a U.S. report, use ÷ 405 with glucose in mg/dL, never ÷ 22.5. And remember the division of labor: insulin describes the mechanism, while glucose and A1C do the diagnosing — which is what AI DiagMe helps you read as a whole.

Sources

Official sources and peer-reviewed publications (PubMed, ClinicalTrials.gov) used for this guide: