By Dr. Muhammad Awais Rabbani, MBBS · Medically reviewed by Dr. Ahmed Humayon, MBBS · Updated 25 October 2026 · Editorial policy
A normal A1C is below 5.7% (39 mmol/mol) for adults without diabetes. 5.7% to 6.4% is prediabetes, and 6.5% or higher on two tests indicates diabetes, per the American Diabetes Association.
| Band | A1C (%) | A1C (mmol/mol) | eAG (mg/dL) |
|---|---|---|---|
| Normal | below 5.7 | below 39 | below 117 |
| Prediabetes | 5.7 to 6.4 | 39 to 47 | 117 to 137 |
| Diabetes | 6.5 or higher | 48 or higher | 140 or higher |
Cut-offs: ADA Standards of Care, Section 2. mmol/mol from the NGSP–IFCC master equation; eAG from the ADAG study.
What is the normal A1C range?
The normal A1C range for an adult without diabetes is below 5.7%, which equals below 39 mmol/mol and an estimated average glucose under 117 mg/dL. An A1C level from 5.7% to 6.4% is prediabetes. A result of 6.5% or higher, confirmed by a second test, meets the diagnostic threshold for diabetes (ADA Standards, Section 2; CDC).
Those three bands are the whole chart. Every A1C chart you find online, including the ones from hospitals and labs, is a rearrangement of the same two cut-offs, 5.7% and 6.5%. What changes between charts is the unit. The US reports a percentage. The UK and most of Europe report mmol/mol, and some reports add an estimated average glucose (eAG) in mg/dL so the number lines up with a finger-stick meter.
The table above shows all three units side by side.
The two unit columns do not end at the same place. The diabetes threshold of 6.5% converts to 48 mmol/mol after rounding, so 47 mmol/mol (about 6.45%) is still prediabetes, even though a US lab would print it as 6.4% or 6.5% (NGSP).
A normal result does not depend on your age or sex for diagnosis. The same cut-offs apply to every non-pregnant adult in the ADA criteria. Pregnancy is the exception, covered in the reliability table below.
Check your A1C result
Enter one of your A1C test results as a percentage or in mmol/mol, and the checker returns three things at once: your band (normal, prediabetes or diabetes range), the value converted to the other unit, and your estimated average glucose in mg/dL from the same reading. It uses the same ADA cut-offs and conversion equations cited on this page.
A1C result checker
Your band, the value in %, mmol/mol and estimated average glucose, and whether the number can be trusted.
Bands: ADA Standards of Care, Section 2 (non-pregnant adults). mmol/mol: NGSP–IFCC master equation. Estimated average glucose: ADAG study (Nathan et al., Diabetes Care 2008). This checker does not diagnose; a clinician interprets your result.
The checker reads a number. It cannot tell you whether that number is trustworthy for your body, which is why the reliability toggle exists.
What does the A1C test measure?
The A1C test measures the percentage of hemoglobin in your red blood cells that has glucose attached to it. Because red blood cells live about 120 days, the result reflects your average blood glucose over roughly the past months, weighted toward the most recent 2 to 3 months (NIDDK). Glucose in the blood sticks to hemoglobin slowly and permanently. The higher your blood sugar levels run, the more hemoglobin gets glycated.
A1C stands for the specific glycated fraction, hemoglobin A1c, which is why the same lab test appears on reports under several names:
- hemoglobin A1C or HbA1C test
- glycated hemoglobin or glycosylated hemoglobin
- HgbA1c or A1c
The erythrocyte lifespan of about 120 days is the reason the A1C test measures a long-term average instead of a single moment. It is also the test’s main weakness. Anything that shortens or lengthens how long your red blood cells survive pulls the number away from your true glucose, which is covered in the reliability section below.
What A1C level is prediabetes?
An A1C level of 5.7% to 6.4% (39 to 46 mmol/mol) is prediabetes. It means your average glucose sits above the normal range but below the threshold for diabetes. The ADA and CDC both use this range to diagnose prediabetes and diabetes risk.
5.7% is a line drawn on a continuous risk curve. A result of 5.6% and a result of 5.8% describe nearly the same average glucose. The cut-off exists so clinicians can act consistently; your risk rises gradually across the whole range.
Within the band, the top end carries more weight. The ADA Standards describe risk of progressing to type 2 diabetes as climbing as A1C rises through 5.7–6.4%. In the Diabetes Prevention Program, a lifestyle program aiming for 7% weight loss cut the risk of developing type 2 diabetes by 58% over an average 2.8 years in adults at high risk (DPP Research Group, NEJM 2002). That kind of weight loss starts with how many calories you need a day.
A prediabetes result usually leads to retesting, often yearly (ADA Standards, Section 2).
What A1C level means diabetes?
An A1C of 6.5% (48 mmol/mol) or higher meets the ADA diagnostic criterion for diabetes. In most cases a single high result is not enough. The ADA Standards require a second abnormal result, either a repeat A1C or a different test, unless symptoms and a very high glucose make the diagnosis clinically obvious.
The confirming test can be another A1C, a fasting plasma glucose (126 mg/dL or higher), or a 2-hour oral glucose tolerance test (OGTT) value of 200 mg/dL or higher (ADA Standards, Section 2). A fasting plasma glucose needs at least 8 hours without food; the OGTT adds a 75 g glucose drink and a second blood draw two hours later. If two different tests disagree, the ADA advises repeating the test that came back abnormal.
One high A1C on a lab report is a reason to retest. Lab error, a hemoglobin variant, or anemia can each push a single result over 6.5%.
Higher values in the diabetes range signal greater risk of diabetes complications over time, but this page stops at the diagnostic line. Very high glucose with too little insulin can also push ketones into the urine, which needs same-day medical advice.
Once a diagnosis is made, the number that matters becomes your personal A1C goal, which depends on age, other conditions and hypoglycemia risk; your clinician sets it with you.
Why your lab report says 4.0–5.6% but guidelines say under 5.7%
Many US labs print a reference interval of about 4.0% to 5.6% next to an A1C result. That interval describes where most healthy people fall statistically. The ADA’s under-5.7% figure is a diagnostic cut-off. They answer different questions, so both can appear in the same discussion without contradicting each other.
A lab’s reference interval comes from testing a healthy population and reporting the range that covers most of them. It is lab-specific: two labs using different NGSP-certified assays may print slightly different lower limits. The diagnostic cut-offs (5.7% and 6.5%) come from outcome studies linking A1C to retinopathy and future diabetes, and they are the same everywhere (ADA Standards, Section 2).
| Lab reference interval | ADA diagnostic cut-off | |
|---|---|---|
| Typical value | ~4.0–5.6% (varies by lab) | below 5.7% normal; 6.5% diabetes |
| Where it comes from | The lab’s healthy-population data | Studies of complications and progression |
| What it answers | “Is this result typical?” | “Does this person have prediabetes or diabetes?” |
| Flags | A “high” flag starts at the lab’s upper limit | Diagnosis needs confirmation |
A point-of-care (POC) A1C, the finger-stick result some clinics produce in minutes, adds a third source of variation. The ADA Standards restrict diagnostic use of POC A1C to FDA-cleared devices run in laboratories that meet CLIA quality standards.
A clinic POC result is fine for tracking; diagnosis should rest on a certified method. A1C is often ordered with a metabolic panel, so it can print beside results such as creatinine levels, each with its own lab-specific reference interval.
A1C in mmol/mol and estimated average glucose
A1C can be reported three ways. US labs use NGSP percentages. Most other countries use IFCC units in mmol/mol. Estimated average glucose (eAG) converts the percentage into mg/dL. The two formulas are IFCC = (NGSP − 2.15) × 10.929 and eAG = 28.7 × A1C − 46.7 (NGSP; Nathan et al., ADAG, Diabetes Care 2008).
The second formula is the ADAG equation, derived from the A1C-Derived Average Glucose study, which compared A1C with frequent glucose readings over three months in people with and without diabetes. It is the standard way to turn a percentage into an average blood sugar figure you can compare with a meter or sensor reading. Some lab reports print eAG directly beneath the percentage for that reason.
| A1C (%) | A1C (mmol/mol) | eAG (mg/dL) |
|---|---|---|
| 5.0 | 31 | 97 |
| 5.5 | 37 | 111 |
| 5.7 | 39 | 117 |
| 6.0 | 42 | 126 |
| 6.4 | 46 | 137 |
| 6.5 | 48 | 140 |
| 7.0 | 53 | 154 |
Calculated from the NGSP–IFCC master equation and the ADAG equation; values rounded.
eAG is an estimate. Two people with the same A1C can have average glucose levels that differ by a meaningful amount, which is one reason the next section exists.
When is an A1C result not reliable?
An A1C result is unreliable whenever red blood cell turnover is abnormal or a hemoglobin variant interferes with the assay. Shorter red-cell survival gives falsely low results; longer survival gives falsely high ones. The ADA Standards say to use plasma glucose criteria alone for diagnosis when red-cell turnover is altered.
These factors affect results in predictable directions. The table gives the direction of error and the test to ask for instead.
| Condition | Direction of error | What to ask for instead |
|---|---|---|
| Iron deficiency anemia | Falsely high (NGSP) | Fasting plasma glucose or OGTT; repeat A1C after iron treatment |
| Hemolysis or recent blood loss | Falsely low (NGSP) | Fasting plasma glucose, OGTT, or fructosamine |
| Recent blood transfusion | Unreliable; often falsely low, because donor cells carry the donor’s A1C (ADA) | Plasma glucose criteria |
| Chronic kidney disease on dialysis | Unreliable; often falsely low (ADA) | Plasma glucose criteria; glucose management indicator (GMI) from a continuous glucose monitor |
| Erythropoietin (ESA) therapy | Falsely low (ADA) | Plasma glucose criteria or GMI |
| Sickle cell disease | Unreliable (ADA) | Plasma glucose criteria; fructosamine |
| Hemoglobin variant or trait (sickle cell trait, HbC, HbE) | Depends on the assay (NIDDK) | An NGSP-certified assay without interference from that variant |
| 2nd/3rd trimester of pregnancy | Lower than true glucose (ADA) | Glucose-based testing |
Variants S and E are more common in people of sub-Saharan African or Mediterranean ancestry and in people of South or Southeast Asian descent (NIDDK). Many carriers have no symptoms and do not know. The NGSP publishes which assay methods each common variant interferes with, so a lab can switch methods.
Fructosamine, a glycated protein test, reflects roughly the past 1 to 2 weeks of glucose and does not depend on red-cell lifespan. Medicare’s coverage policy calls it useful in patients with hemolytic anemia or hemoglobinopathies (CMS NCD 190.21). GMI, calculated from at least 14 days of sensor data, gives an A1C-like number straight from glucose readings (Bergenstal et al., Diabetes Care 2018).
Suspect an unreliable result when a normal A1C sits next to high fasting or meter readings, or the reverse. When the two disagree, the glucose readings usually deserve more trust.
An unusually low A1C needs the same scrutiny, because blood loss and hemolysis are the usual causes. In kidney disease, kidney function itself is tracked with blood creatinine and eGFR, not with A1C.
Do you need to fast for an A1C test?
No. The A1C test does not require fasting, because it reflects average glucose over about three months and is barely affected by a recent meal. That makes it more convenient than a fasting plasma glucose, which needs at least 8 hours without food (NIDDK; ADA Standards).
It is a simple blood test. The blood sample comes either from a vein, sent to a lab, or from a finger-stick on a POC device. Either way, the draw takes a few minutes and needs no special preparation. Ask before the appointment whether other blood glucose tests are ordered at the same visit, since those may still need fasting.
How often the test is repeated depends on the result:
- Normal result with risk factors: the ADA recommends screening at least every 3 years (ADA Standards).
- Prediabetes: at least once a year (ADA Standards, Section 2).
- Diagnosed diabetes: at least twice a year when on target, every 3 months after a treatment change or when above goal (ADA Standards, Section 6). Your health care professional sets the schedule as part of your treatment plan and diabetes management.
Is the A1C test covered? Billing codes and frequency
Most US insurers cover an A1C when it meets medical necessity rules. Coverage hinges on the code pairing: CPT 83036 for a lab-run A1C, CPT 83037 for an FDA-cleared home-use device at point of care, matched with a diagnosis code such as Z13.1 (screening) or R73.03 (prediabetes). Medicare began covering A1C for diabetes screening on January 1, 2024 (CMS MM13487).
Dr. Rabbani also runs Growth Partners Global, a US medical billing company; the codes and rules below come from published CPT and CMS guidance.
| Code | What it is | When it is used |
|---|---|---|
| CPT 83036 | Hemoglobin; glycosylated (A1C), lab method | Most venous lab tests (AMA CPT) |
| CPT 83037 | Hemoglobin; glycosylated (A1C) by device cleared by FDA for home use | Clinic point-of-care testing (AMA CPT) |
| ICD-10-CM Z13.1 | Encounter for screening for diabetes mellitus | Screening in people without a diagnosis (CMS ICD-10) |
| ICD-10-CM R73.03 | Prediabetes | Follow-up testing after a prediabetes result (CMS ICD-10) |
For Medicare diabetes screening, CMS allows up to two screening tests within a 12-month period and waives coinsurance and the Part B deductible, because the USPSTF diabetes screening recommendation includes A1C (CMS MM13487; 42 CFR 410.18). After a diabetes diagnosis, A1C moves to the monitoring rules: Medicare treats testing more often than every three months in a controlled patient as not reasonable and necessary (CMS NCD 190.21).
Is a “good” A1C the same for everyone?
For diagnosis, yes: below 5.7% is normal for every non-pregnant adult under ADA criteria. For people already diagnosed, no. A “good” A1C then becomes an individual target set by a clinician, and it shifts with age, pregnancy, hypoglycemia risk and other health conditions (ADA Standards).
This page explains A1C ranges and test results; it does not replace a clinician’s interpretation of your own blood test results.