An HbA1c blood test reports what share of your hemoglobin is carrying glucose. Every guide we write starts from the panel a reader already has. This one sits on almost all of them, printed as a bare percentage with no explanation next to it.
The Verdict
What an HbA1c Blood Test Physically Measures
Glucose in the blood attaches to hemoglobin inside red blood cells, slowly and without any enzyme involved. The higher the glucose concentration and the longer the exposure, the more hemoglobin ends up glycated. HbA1c is the percentage of total hemoglobin that has undergone this reaction.
That mechanism explains the properties people find confusing. The reaction is irreversible for the life of the cell, so the number cannot be lowered by a day of good behaviour. It is also a proportion rather than a concentration, which is why anything that changes red cell turnover changes the result without any change in your blood sugar.
Results are standardised through the National Glycohemoglobin Standardization Program, which is why a percentage from one United States laboratory is comparable to one from another. Outside the United States you may see the same test reported in mmol/mol under the International Federation of Clinical Chemistry scale. A 6.5% result is 48 mmol/mol. The blood is identical, only the reporting units differ.
The Bands, and Where They Came From
The three HbA1c bands are diagnostic cutpoints drawn from population risk data. They are not points where the body changes state.
| HbA1c | Label | What follows |
|---|---|---|
| Below 5.7% | Normal | Nothing to act on from this number alone. Rescreen every three years from age 35, sooner with risk factors. |
| 5.7% to 6.4% | Prediabetes | A range rather than a diagnosis of disease. The Centers for Disease Control and Prevention (CDC) puts progression to type 2 diabetes at 15% to 30% within five years without a change in diet, weight or activity. |
| 6.5% and above | Diabetes threshold | Diagnosis needs two abnormal results, or one result plus symptoms. A single high reading is a reason to repeat the test rather than to start treatment. |
The 6.5% threshold came from the risk of diabetic retinopathy, which climbs steeply above that value. An International Expert Committee report in Diabetes Care set it on that basis in 2009. The line therefore marks where eye damage becomes common. It does not mark where glucose handling first goes wrong. Someone at 6.4% is not meaningfully different from someone at 6.5%, and the label attached to them is.
The lower cutpoint carries the same caveat in the other direction. A result of 5.6% falls inside the normal band and sits close enough to the prediabetes line that a repeat in twelve months is reasonable. We treat the bands as a rough map, and read them next to fasting glucose, waist measurement and fasting insulin rather than on their own.
Why You Do Not Fast for an HbA1c Test
Fasting changes nothing on this test, because the glycation being measured happened weeks ago. Breakfast, coffee with sugar, or a large meal the night before all leave the result where it was.
Readers still arrive fasted, and usually for a good reason: the HbA1c was ordered alongside a lipid panel or a fasting glucose on the same form, and those tests do have preparation rules. Check the requisition rather than the marker. If HbA1c is the only test listed, book the appointment that fits your day. Our page on fasting for a lipid panel covers the tests on that same form that genuinely need it.
How the Three-Month Window Is Weighted
HbA1c is not an even average across 90 days. Red cells are constantly being replaced, so the youngest cells in the sample have had the least exposure and the oldest have had the most. The arithmetic works out at roughly 50% of the value coming from the most recent 30 days, about 25% from days 30 to 60, and the remainder from the two months before that.
Two practical consequences follow. A person who tightens up their diet four weeks before a scheduled draw sees a number better than their real three-month average. That is a common reason a result improves and then rebounds at the next test. Someone who genuinely changed their habits four months ago is already seeing nearly the full effect, so waiting longer for a "truer" reading gains nothing.
Assay imprecision sets the other limit on what a change means. Two samples from the same blood can differ by a few hundredths of a percent, and repeat measurements in the same person over a short period commonly vary by up to about 0.3%. A move from 5.8% to 5.6% is inside that noise. A move from 6.4% to 5.8% is not.
Estimated Average Glucose, and Converting the Percentage
Most laboratory reports print an estimated average glucose, or eAG, next to the percentage. It converts HbA1c into the units a glucose meter shows, using the formula 28.7 times HbA1c minus 46.7 for a result in mg/dL.
| HbA1c | Estimated average glucose | How to read it |
|---|---|---|
| 5.0% | 97 mg/dL | Comfortably normal |
| 5.7% | 117 mg/dL | The lower prediabetes cutpoint |
| 6.0% | 126 mg/dL | Mid-prediabetes |
| 6.5% | 140 mg/dL | The diagnostic threshold |
| 7.0% | 154 mg/dL | The treatment target most guidelines set for adults with diabetes |
The conversion is useful for translating between a laboratory result and a meter or sensor reading, and it carries one warning. An average conceals the shape of the curve underneath it. A person whose glucose runs flat at 117 mg/dL and one who spikes to 190 mg/dL after every meal and drops to 75 mg/dL overnight can produce the same 5.7%. Only a continuous glucose monitor separates them, and our page on how accurate CGMs are covers how closely a sensor average tracks a laboratory HbA1c.
What Moves an HbA1c Result Without Any Change in Glucose
Because HbA1c is a proportion of hemoglobin, anything that shortens or lengthens red cell survival moves the number on its own. The list is short and each item is common.
- Iron deficiency raises it. Older cells persist when production is limited, so more of the hemoglobin in the sample has had a long exposure. Correcting the deficiency lowers HbA1c with no change in diet.
- Recent blood loss or a transfusion lowers it. New cells replace old ones and reset the clock on part of the sample.
- Chronic kidney disease pulls it down through shortened red cell survival, which is why HbA1c under-reads in advanced kidney disease.
- Hemoglobin variants interfere with some assays. HbS, HbC, HbE and HbD trait can shift results on certain methods while leaving others untouched. The NGSP maintains a table of which methods are affected by which variant.
- Pregnancy lowers it through faster red cell turnover, which is one reason gestational diabetes is diagnosed on an oral glucose tolerance test instead.
Any of these makes the percentage describe your red cells rather than your glucose. Our page on a falsely high or low A1c works through each case and names the test to order in its place.
Who Should Not Rely on HbA1c Alone
Anyone in the five situations above should treat a single HbA1c as provisional and ask for a fasting glucose or an oral glucose tolerance test alongside it. A fructosamine test, which reflects the previous two to three weeks and does not involve hemoglobin, is the usual alternative when red cell turnover is the problem.
The other group is people asking a question HbA1c was never built to answer. If you want to know whether insulin resistance is developing, this is the wrong marker, because the pancreas compensates for years before glucose rises at all. Fasting insulin, HOMA-IR and the triglyceride to HDL ratio all move earlier. A normal HbA1c is not a clearance on metabolic health.
What would change our answer is a cheaper, standardised insulin assay. If fasting insulin were reported comparably across laboratories the way HbA1c is, we would recommend it as the first-line metabolic screen and leave HbA1c to its diagnostic job. Until that happens, HbA1c is what nearly every panel includes and the number most readers actually have in front of them.
When a Result Warrants Seeing a Physician
A first result at 6.5% or higher needs a clinician, and it needs confirmation before anything else. Diagnosis normally requires two abnormal results from separate draws, or one abnormal result together with symptoms such as unexplained thirst, frequent urination or weight loss.
A result in the 5.7% to 6.4% band is a prompt rather than an emergency, and the response that has trial evidence behind it is a structured lifestyle programme. The CDC-recognised National Diabetes Prevention Program is the version tested at scale. Our guide to reversing prediabetes covers what the trials actually achieved and how long it took.
Book the test without fasting, ask for a fasting glucose on the same draw, and bring both numbers to your next appointment rather than acting on the percentage alone.
Frequently Asked Questions
Do you need to fast for a hemoglobin A1c test?
No. HbA1c measures hemoglobin that has already been glycated over the previous months, so nothing you ate that morning changes it. You can eat, drink coffee and take your usual medicines before the draw. The reason people fast for these appointments is that HbA1c is often ordered on the same requisition as a lipid panel or a fasting glucose, and those two do have fasting rules. If HbA1c is the only test on the form, book whatever appointment slot suits you.
What is a normal hemoglobin A1c level?
Below 5.7% is the normal band under the American Diabetes Association criteria, 5.7% to 6.4% is prediabetes, and 6.5% or higher meets the diagnostic threshold for type 2 diabetes. Those cutpoints were set from the risk of diabetic retinopathy in population data rather than from any point where metabolism visibly changes. Our guide to normal versus optimal ranges covers why a result of 5.6% is inside the normal band and still worth watching.
How long does it take for HbA1c to change?
Give it three months. Red blood cells live about 120 days, and the test reads the glucose bound to the hemoglobin inside them, so the number reflects roughly that window. The weighting is not even across it. Around half the value comes from the most recent 30 days and only about a quarter from days 60 to 120. A strict month before a draw therefore moves the number more than the two months before it. Retesting after six weeks tells you very little.
What is estimated average glucose on an HbA1c report?
Estimated average glucose, printed as eAG, converts the percentage into the glucose units a meter shows. The formula is 28.7 times your HbA1c minus 46.7, giving mg/dL. An HbA1c of 6.0% works out at 126 mg/dL. It is an average across the whole period, so it hides both the overnight lows and the post-meal peaks that a continuous glucose monitor would show. Two people with identical eAG can have completely different glucose curves.
Can a hemoglobin A1c test be wrong?
Yes, and the causes are common enough to check before acting on a surprising result. Iron deficiency raises HbA1c without any change in glucose. Recent blood loss, a transfusion, pregnancy and chronic kidney disease all pull it down. Hemoglobin variants such as HbS, HbC and HbE interfere with some assay methods and not others. We cover each of these, and what to order instead, on when an A1c result is wrong.
How often should HbA1c be tested?
For screening in an adult with no diagnosis, the American Diabetes Association suggests starting at age 35 and repeating every three years if the result is normal. Anyone in the prediabetes band is usually retested yearly. People treated for diabetes are tested twice a year when results are stable and at target, and quarterly when treatment has changed or the target is being missed. Testing more often than quarterly rarely produces a number that means anything new, because the marker itself cannot move that fast.
Is HbA1c better than a fasting glucose test?
It answers a different question. Fasting glucose is one moment, sensitive to a bad night of sleep, a late meal or a stressful morning. HbA1c averages months and is unaffected by any of that. The tradeoff is speed: HbA1c is slow to register a real improvement and can stay in the prediabetes band for a quarter after glucose has already normalised. Ordering both on the same draw costs little and resolves most disagreements between them.
What does the HbA1c test not tell you?
It cannot see variability, and that is its main blind spot. A person spiking to 180 mg/dL after meals and dropping to 70 mg/dL overnight can return the same HbA1c as someone sitting flat at 110 mg/dL all day. It also says nothing about insulin. Insulin resistance runs for years with a normal HbA1c while the pancreas compensates, which is why fasting insulin and the triglyceride to HDL ratio catch the problem earlier.