HbA1c: what the number means

Reviewed 16 sourcesPublished by PepVault

What your HbA1c measures, the ADA's normal, prediabetes and diabetes cut-offs in % and mmol/mol, the average glucose behind it and what makes it read wrong.

Silver-haired woman in glasses and a periwinkle cardigan smiles at a blank sheet in a kitchen
Last reviewed
Sources
16
Questions
6
Compounds
2

The short answer

HbA1c is the share of your hemoglobin with glucose stuck to it, an average of the last two to three months that leans on the most recent weeks. The ADA calls below 5.7% (39 mmol/mol) normal, 5.7 to 6.4% prediabetes and 6.5% (48 mmol/mol) or more diabetes, confirmed by a second test. A 7% means an average glucose of about 154 mg/dL.

  • ADA 2026: below 5.7% (39 mmol/mol) is normal, 5.7 to 6.4% (39 to 47 mmol/mol) is prediabetes, 6.5% (48 mmol/mol) or more is diabetes, confirmed by a second test.
  • The average glucose behind a number is wide: a 7% matched an average of 123 to 185 mg/dL in the study that set the formula.
  • Iron deficiency can push HbA1c up; blood loss, hemolysis, pregnancy, dialysis and some hemoglobin variants can pull it down or make it unreliable.
  • In the label trials, GLP-1s cut HbA1c by 1.4 to 2.3 points from about 8% in type 2 diabetes, and by 0.4 points (0.3 to 0.4 more than placebo) in people without diabetes.

Every number checked at its primary source · · 16 sources · How we research

Evidence: Strong: the cut-offs are the ADA's 2026 Standards of Care, the glucose conversion comes from a 507-person study with about 2,700 readings each, and the GLP-1 numbers are randomized trial results printed in FDA labels; the interference data on non-iron anemias are limited.

Silver-haired woman in glasses and a periwinkle cardigan smiles at a blank sheet in a kitchen

What does HbA1c actually measure?

Hemoglobin is the protein in red blood cells that carries oxygen. Glucose in the blood sticks to it slowly and stays stuck, a process called glycation, and HbA1c (also written A1C) is the percentage of hemoglobin that is glycated. The more glucose around, the higher the number. A red cell lives about 120 days, so the test reads glucose over roughly the past two to three months15.

It is not a flat average. Every red cell in the sample counts, from the newest to the oldest, so the most recent 30 days weigh more than the days 90 to 120 back5. When nine people with newly diagnosed type 2 diabetes had their glucose brought to normal within days, HbA1c fell by half in about 35 days, and the weighting reached back roughly 100 days8. A big change in glucose therefore shows up in HbA1c within weeks, but the full effect takes about three months.

Because it is a long average, HbA1c is not moved by what you ate that morning, by stress or by a short illness, and you do not need to fast for it1.

A smiling woman in a blue top leaning on her kitchen counter with her phone and a notebook

PepVault for iPhone

Keep your protocol in PepVault

PepVault's bloodwork log keeps each HbA1c with its panel date, flags it against a reference range you can set to your lab's, and adds it to that marker's trend, so each three-month result sits beside the last.

Download on the App Store

7 days free on the yearly plan for new subscribers.

What HbA1c is normal, prediabetes or diabetes?

The cut-offs below are the American Diabetes Association's (ADA) 2026 Standards of Care. Labs outside the US often report HbA1c in mmol/mol (the IFCC unit, millimoles of glycated hemoglobin per mole of hemoglobin) rather than percent; both describe the same test16.

ADA category1HbA1c (%)HbA1c (mmol/mol)
NormalBelow 5.7Below 39
Prediabetes5.7 to 6.439 to 47
Diabetes6.5 or higher48 or higher

One high result is not a diagnosis on its own. Unless glucose is unmistakably high, the ADA requires a confirming result: a different test, such as fasting glucose, which can come from the same blood draw, or the same test repeated on another day1. The lab's method should be certified by the NGSP (the US program that ties every lab's HbA1c to the reference used in the big diabetes trials), and point-of-care devices (a finger-stick test in the clinic) need FDA clearance and a certified site1. Even certified tests vary: NIDDK gives the example of a 6.8% that could read anywhere from 6.4 to 7.2% on another run11.

Once someone has diabetes, the number becomes a goal rather than a cut-off: the ADA calls under 7% (53 mmol/mol) appropriate for many non-pregnant adults, with stricter or looser goals set person by person2. Your clinician reads your own result against your history and your other tests.

When to get help An HbA1c of 6.5% or more, or new thirst, frequent urination and blurred vision, is a reason to book a visit with your clinician soon116.

What average blood sugar does your HbA1c mean?

The conversion comes from the ADAG study, which gave 507 adults (268 with type 1 diabetes, 159 with type 2 and 80 without diabetes) about 2,700 glucose readings each over three months, from sensors and finger sticks, then compared them with a lab HbA1c. The line that fit best was: estimated average glucose (eAG) in mg/dL = 28.7 × HbA1c − 46.7. It held across age, sex, diabetes type and ethnicity4.

HbA1c (%)HbA1c (mmol/mol)Average glucose (mg/dL)Average glucose (mmol/L)Range for 95% of people (mg/dL)
5.031975.476 to 1202
5.7391176.5Not published
6.0421267.0100 to 1522
6.5481407.8Not published
7.0531548.6123 to 1852
8.06418310.2147 to 2172
9.07521211.8170 to 2492
10.08624013.4193 to 2822

The mg/dL column is the ADAG formula rounded; mmol/L is mg/dL divided by 18, the same as the ADAG paper's own 1.59 × HbA1c − 2.59; mmol/mol comes from the NGSP's master equation, HbA1c % = 0.09148 × mmol/mol + 2.152467. The whole-number rows match the ADA's own table2.

The last column is the part most pages leave out. Two people with the same 7% can run very different averages, about 123 to 185 mg/dL, because the link between glucose and HbA1c differs from person to person2. That is also why the estimated A1C a glucose sensor app shows often differs from the lab number: our guide to CGMs without diabetes covers that gap.

What makes HbA1c read falsely high or low?

Anything that changes how long red cells live, or how the lab's method sees hemoglobin, changes the number without any change in glucose. The ADA lists these and says glucose tests, not HbA1c, should be used to diagnose diabetes when one applies1.

ConditionEffect on HbA1cWhy
Iron deficiency, with or without anemiaHigher91011Not fully settled; one lead is a by-product (malondialdehyde) that speeds glycation. Iron treatment brings HbA1c back down9
Recovery from blood loss, hemolytic anemiaLower9Younger red cells have had less time to collect glucose
Other anemias (not iron deficiency)Possibly lower; few studies10Altered red cell turnover
Transfusion, erythropoietin treatment, splenectomyChanged, direction varies111Red cell age shifts
Kidney failure and dialysisTends to read low on dialysis9Kidney anemia, erythropoietin, and a modified hemoglobin that some methods misread9
PregnancySlightly lower3Faster red cell turnover; late pregnancy can rise again with iron deficiency9
Hemoglobin variants (sickle cell trait, HbC, HbE, HbD) and high fetal hemoglobinDepends on the lab's method9Some methods misread the variant; others are unaffected
G6PD deficiency (the inherited G202A variant, two copies)About 0.7 to 0.8 points lower1Faster red cell turnover1
Severe high triglycerides or bilirubinDepends on the method1Interferes with some assays

NIDDK flags HbA1c as possibly unreliable for people of African, Mediterranean or Southeast Asian descent who carry a variant, and the ADA asks for a method without variant interference, for example in sickle cell trait111. Cirrhosis and HIV are on the ADA's list too1. If your HbA1c and your glucose readings do not agree, these are the usual reasons, and a fasting glucose, a glucose tolerance test or fructosamine can stand in12.

How often should HbA1c be tested?

The ADA's schedule depends on where you start123:

  • No diabetes and a normal result: screening starts at 35, earlier with risk factors such as overweight plus a family history, and repeats at least every 3 years1.
  • Prediabetes: once a year1.
  • Diabetes and at goal: at least twice a year2.
  • Not at goal, or a recent change in treatment (for example starting a GLP-1): about every 3 months2.
  • Pregnancy with diabetes: as often as monthly, as a second measure after glucose monitoring3.

The three-month rhythm follows from the biology above: a recheck at four weeks catches only part of a change, because about half of the move shows up in the first five weeks8.

PepVault's bloodwork log keeps each HbA1c with its panel date, flags it against a reference range you can set to your lab's, and adds it to that marker's trend, so each three-month result sits beside the last. Log it in PepVault

How much do GLP-1 drugs lower HbA1c?

HbA1c is the main result in every diabetes trial of these drugs, and the FDA labels print it. The doses below are the trials' doses, told as trial results12131415.

Trial (label)WhoWeeksStartChange on drugComparator, and the label's difference
SUSTAIN 1, Ozempic12388 adults with type 2 diabetes308.0 to 8.1%−1.4 (0.5 mg), −1.6 (1 mg)−0.1 (placebo); difference −1.2 and −1.4
SURPASS-1, Mounjaro13478 adults with type 2 diabetes407.9 to 8.1%−1.8 (5 mg), −1.7 (10 mg), −1.7 (15 mg)−0.1 (placebo); difference −1.7, −1.6, −1.6
SURPASS-2, Mounjaro131,879 adults with type 2 diabetes on metformin408.3%−2.0 (5 mg), −2.2 (10 mg), −2.3 (15 mg)−1.9 (semaglutide 1 mg); difference −0.2, −0.4, −0.5
STEP 2, Wegovy 2.4 mg (label Study 3)14807 adults with type 2 diabetes and overweight688.1%−1.6−0.4 (placebo); difference −1.2
SURMOUNT-2, Zepbound (label Study 2)15938 adults with type 2 diabetes and overweight728.0 to 8.1%−2.1 (10 mg and 15 mg)−0.5 (placebo); difference −1.6
STEP 1, Wegovy 2.4 mg (label Study 2)141,961 adults without diabetes685.7%−0.4−0.2 (placebo); difference −0.3
SURMOUNT-1, Zepbound (label Study 1)152,539 adults without diabetes725.5 to 5.6%−0.4 (each dose)−0.1 (placebo); difference −0.3 (5 mg), −0.4 (10 and 15 mg)

The pattern is plain: the higher the starting number, the bigger the fall. From about 8%, these drugs took HbA1c down by 1.4 to 2.3 points, and 70 to 86% of people on them ended under 7%, against 23 to 28% on placebo1213. The "change" is each group's own fall; the label's difference is the drug's fall minus the comparator's. In people without diabetes, starting near 5.6%, the labels put the fall at 0.3 to 0.4 points more than placebo1415, which is the size of a move out of the prediabetes band for someone sitting just inside it. For metformin and berberine on the same scale, see our comparison guide.

Commonly reported: a first recheck a few weeks after starting a GLP-1 that has moved less than hoped, and a clear drop at the three-month test; and a lab HbA1c that differs from the estimate in a glucose sensor app.

HbA1c says nothing about lows or swings within the day, and it is one marker among several. If you are working out whether insulin resistance sits behind a normal HbA1c, our insulin resistance tests guide covers fasting insulin, HOMA-IR and the glucose tolerance test.

Track it in the app

Sources

  1. American Diabetes Association Professional Practice Committee. 2. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes—2026. Diabetes Care. 2026;49(Suppl 1):S27–S49. PubMed 41358893
  2. American Diabetes Association Professional Practice Committee. 6. Glycemic Goals, Hypoglycemia, and Hyperglycemic Crises: Standards of Care in Diabetes—2026. Diabetes Care. 2026;49(Suppl 1):S132–S149 (Table 6.1). PubMed 41358894
  3. American Diabetes Association Professional Practice Committee. 15. Management of Diabetes in Pregnancy: Standards of Care in Diabetes—2026. Diabetes Care. 2026;49(Suppl 1):S321–S338. PubMed 41358885
  4. Nathan DM et al. Translating the A1C assay into estimated average glucose values. Diabetes Care. 2008. PubMed 18540046
  5. National Glycohemoglobin Standardization Program (NGSP). HbA1c and estimated average glucose (eAG). ngsp.org
  6. National Glycohemoglobin Standardization Program (NGSP). IFCC standardization of HbA1c: the IFCC and NGSP (master equation and Table 1). ngsp.org
  7. Hoelzel W et al. IFCC reference system for measurement of hemoglobin A1c in human blood and the national standardization schemes in the United States, Japan, and Sweden: a method-comparison study. Clin Chem. 2004. PubMed 14709644
  8. Tahara Y, Shima K. Kinetics of HbA1c, glycated albumin, and fructosamine and analysis of their weight functions against preceding plasma glucose level. Diabetes Care. 1995. PubMed 7497851
  9. National Glycohemoglobin Standardization Program (NGSP). Factors that interfere with HbA1c test results (updated October 23, 2026). ngsp.org
  10. English E et al. The effect of anaemia and abnormalities of erythrocyte indices on HbA1c analysis: a systematic review. Diabetologia. 2015. PubMed 25994072
  11. National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). The A1C test and diabetes. niddk.nih.gov
  12. Ozempic (semaglutide) injection prescribing information, section 14 (Table 3). Novo Nordisk; DailyMed, effective June 2026. dailymed.nlm.nih.gov
  13. Mounjaro (tirzepatide) prescribing information, sections 14.2 and 14.3 (Tables 4 and 5). Eli Lilly; DailyMed, effective August 2026. dailymed.nlm.nih.gov
  14. Wegovy (semaglutide) prescribing information, section 14 (Table 13). Novo Nordisk; DailyMed, effective June 2026. dailymed.nlm.nih.gov
  15. Zepbound (tirzepatide) prescribing information, section 14 (Table 3). Eli Lilly; DailyMed, effective August 2026. dailymed.nlm.nih.gov
  16. National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). Symptoms and causes of diabetes. niddk.nih.gov

Newsletter · Weekly

The weekly peptide briefing.

New guides, the week’s peptide news, one tool. Free, one email a week, unsubscribe anytime.

We’ll email you a link to confirm. Your address is used for the briefing only. Privacy

PepVault for iPhone

Keep your protocol in PepVault

Log every dose, track every vial, and see your protocol come together.

Download on the App Store

7 days free on the yearly plan for new subscribers.

Explore the live demo