On This Page – Quick Medical Summary
How accurate is the A1C test? The short answer
Yes — accurate enough to diagnose and manage diabetes with, and more stable than the fasting glucose tests it competes with. Before any US laboratory may report your A1C test result for a diagnosis, the method it runs has to pass a documented comparison against a reference laboratory, and the lab itself has to be graded on unknown samples several times a year.
The limits are real. They are also known, measured and mostly checkable — which is the part almost nobody explains.
Where to go next depends on why you are asking:
- A result came back higher than you expected. Start with what the lab had to prove.
- Your A1C disagrees with your meter, your CGM or your last test. Go to what can make a result wrong for you, then to the three things you can do.
- You’ve been told the test doesn’t work for people with your family background. That warning is older than the evidence.
- You’re reading for a parent, partner or someone you care for. The question to bring to their appointment is in the section on checking a result.
For what the number itself means, start instead with how to read your A1C result and its honest margin.
ℹ️ Medical Disclaimer: This article is general education about how a laboratory test is measured and checked. It is not a diagnosis, not an interpretation of your own result, and not a basis for starting, stopping or changing any medication, treatment or monitoring plan. Decisions about diagnosis, retesting, treatment and insurance coverage belong to a licensed clinician who has your full history in front of them. If a result concerns you, take it to a board-certified primary care physician or endocrinologist rather than acting on anything here.
What your lab had to prove before reporting your A1C
There is no single “A1C test.” There is a family of roughly forty commercial assay methods, each one certified separately, and knowing that is what makes everything below make sense.
When an A1C is used for diagnosis, the sample is drawn from a vein and sent to a laboratory running an NGSP-certified method. As the federal patient guide to the A1C test, last reviewed in April 2018, explains, that certification exists so different makers’ tests produce results comparable with the landmark trial the modern thresholds came from — the reason the A1C bands mean the same thing everywhere.
What “NGSP-certified” actually means
A manufacturer sends forty patient samples to a secondary reference laboratory, both run them, and the results are compared one by one. To pass, at least thirty-six of the forty must land within 5% of the reference figure; a lab seeking Level I certification has to hit thirty-seven of forty.
Everything traces to one anchor — the reference method used in the Diabetes Control and Complications Trial — which is why a 7.0% in Ohio means what a 7.0% means in Oregon.
Why the certificate is specific to one instrument and one set of reagents
A certificate is valid for one year and covers one named instrument running one named set of reagents. Change either and the certification does not travel with it.
That sounds like fine print. It is the most useful fact here, and the section on checking your own result explains why.
What proficiency testing is, and how often it happens
Certification covers the method. Proficiency testing covers the laboratory actually running it.
Under federal rules, labs doing moderate- or high-complexity testing must enroll in an approved program that mails them unknown samples, which they run exactly as they would run yours. They report the results and get graded. In the words of the agency that sets the rule, this independently verifies the accuracy and reliability of laboratory testing, including the competence of the people doing it — and for A1C it happens three times a year.
Current guidance treats A1C alongside continuous glucose monitoring metrics as the standard way of assessing glycemic status — see the 2026 Standards of Care section on glycemic goals.
🔬 How It Works: Glucose in your bloodstream attaches to hemoglobin inside red blood cells and stays attached for the life of the cell. The higher your glucose has been, the more hemoglobin carries sugar. An A1C method measures what proportion is glycated — which is why anything that changes how long your red blood cells live changes the answer, no matter how good the machine is.
Before those distortions, one question: are this system’s tolerances unusual? They are not.
How A1C compares with other blood tests for accuracy
Three American bodies set three different tolerances for A1C, and no consumer health page prints any of them.
One definition first, because it changes how you read the numbers. In the federal rules, an acceptance limit is the symmetrical tolerance — plus and minus — around a target value, and it is the outer bound used to grade a laboratory’s performance on unknown test samples. It is not an estimate of how wrong your own result is, and typical method performance sits well inside it.
The three different tolerances, and who sets each
| Who sets it | What it governs | Limit for A1C | Key clinical detail |
|---|---|---|---|
| NGSP | Certifying that a method may be used at all | 36 of 40 samples within ±5% of a reference lab | The tightest of the three, and the bar a method must clear before it reaches any patient |
| College of American Pathologists | Its own accreditation checklist for accuracy-based testing | ±6% | Stricter than the federal floor; applies to CAP-accredited labs |
| Federal CLIA regulation | Grading a lab’s proficiency-testing performance | Target value ±8% | The regulatory floor every lab must clear, not the expected error on your result |
Sources: NGSP Summary of Certification Criteria; the federal table of acceptable performance limits for laboratory tests, Attachment 1 (8 July 2024); College of American Pathologists checklist criteria.
Now the comparison that answers the question. These figures come from the same federal table, in the same section, graded the same way:
| Blood test | Federal acceptance limit | Key clinical detail |
|---|---|---|
| Hemoglobin (from a blood count) | Target ±4% | The tightest on the routine chemistry list |
| Hemoglobin A1C | Target ±8% | Graded more tightly than most chemistry on the same report |
| Total cholesterol | Target ±10% | |
| LDL cholesterol, direct measurement | Target ±20% | Two and a half times A1C’s tolerance |
| Thyroid-stimulating hormone | Target ±20% |
Source: CMS memorandum QSO-24-15-CLIA, Attachment 1, §493.931 routine chemistry (8 July 2024).
Why A1C only became a federally graded test in 2024
A1C was not a regulated analyte under the federal proficiency-testing rules until very recently. Those rules were written in 1992, and the agency’s explanation is blunt: tests like hemoglobin A1c were not routinely performed before then.
A1C was one of twenty-nine analytes added by a final rule published in July 2022, whose proficiency-testing provisions took effect in July 2024 — the first update to those rules since they were created.
📊 Clinical Data Point: A1C’s federal acceptance limit is target value ±8%, against ±10% for total cholesterol, ±20% for directly measured LDL and ±20% for TSH — Source: CMS, QSO-24-15-CLIA Attachment 1, July 2024.
So the honest answer to “how much variation is allowed?” is: less than for most of the other numbers on your report — under a system that only recently got around to grading this test at all.
What that system cannot control is you.
What can make your A1C wrong for you
The A1C reflects two things: your average glucose, and how long your red blood cells live. Anything that changes the second changes the result, however good the machine is.
Conditions that shorten the life of a red blood cell
This is the one direction of interference that holds regardless of method. Any condition shortening red blood cell survival — recovery from acute blood loss, or a hemolytic anemia — falsely lowers the A1C.
The federal patient guide lists the same family: recent blood loss, sickle cell disease, erythropoietin treatment, hemodialysis or transfusion can all change A1C levels, and kidney failure or liver disease can produce false results too.
A falsely low result is the one worth understanding, because it leaves a problem looking smaller than it is — and it is why a normal A1C rules out less than people assume. If any of that describes your last few months, what low hemoglobin means on a blood count covers the blood-count side.
Low iron, and why it pushes the result the other way
Iron-deficiency anemia goes the other way. It is associated with a higher A1C, and treating the deficiency brings the figure back down — documented in the standardization program’s interference guidance and in the federal patient guide, which states plainly that a falsely high A1C can occur in people very low in iron.
If iron has come up in your own results, how iron studies are read explains those numbers alongside this one.
Kidney disease, where the evidence is genuinely unsettled
Most health sites give you a tidy two-column list of things that make A1C “falsely low” and things that make it “falsely high.” The sources do not support that split for most items, and kidney disease is the clearest example.
The standardization program’s guidance calls the role of A1C in chronic renal failure controversial, notes that recent reports suggest it underestimates glucose control in people on dialysis, and says further studies are needed. That is where the evidence stands. We would rather tell you that than hand you a cleaner list we cannot source.
The tell: when your A1C and your glucose don’t agree
One signal matters more than any list. Clinicians suspect interference — a falsely high or low result — when your A1C and your blood glucose results don’t match. The coordinator of the national A1C standardization network puts it the same way: whenever there is a discrepancy between an A1C and the clinical impression from other tests, the accuracy of that result should be verified.
A mismatch is a reason to ask a question, not to discount the result — which is exactly why a diagnosis is confirmed with a second measurement. If the glucose side is what you are unsure about, what a normal blood sugar reading means covers it, and our blood sugar converter moves a reading between mg/dL and mmol/L.
✅ Patient Action: If your A1C and your glucose readings don’t line up, bring one sentence to your primary care physician: “My A1C and my glucose readings don’t match — could something be affecting the A1C measurement itself?” That question, not a list of conditions, is what gets the right test ordered.
Everything above is something a clinician evaluates alongside your history — none of it is self-assessable. And one widely repeated version of it is now out of date.
The hemoglobin variant warning, brought up to date
If you are of African, Mediterranean or Southeast Asian descent, you have probably been told the A1C may not be reliable for you. That warning was accurate when written. The evidence has moved.
What a hemoglobin variant is, and how common it is
About 98% of normal human hemoglobin is hemoglobin A. Some people carry a second kind too — hemoglobin S, C, D or E. These are hemoglobin variants; in worldwide prevalence sickle cell trait comes first, then E, C and D.
📊 Clinical Data Point: About 7% of the world’s population carries some heterozygous hemoglobin variant — Source: NIDDK, Interpreting A1C: Diabetes and Hemoglobin Variants, February 2020.
Why most carriers never find out
Carrying the trait is clinically silent: less than half the hemoglobin is the variant form, there is no disease, and no reason to test for it — so most carriers never know. The federal patient guide agrees that most people with a variant have no symptoms and may not know they carry one. Some are picked up in pregnancy screening; more recently, some during A1C testing itself.
What the current method evidence shows
The standardization program publishes a table of the twenty methods most often used to measure A1C, showing whether each is affected by hemoglobin C, S, E or D trait, or by elevated fetal hemoglobin, against a defined threshold for clinically significant interference.
In the table as updated in June 2026, for those four common traits, nearly every entry reads “no interference.” One carries a qualified yes, with conflicting data in the literature.
Set that against November 2007, when the National Institutes of Health announced a campaign for exactly this problem: more than 3,000 US laboratories were using twenty different methods, and six of them gave unreliable results in people with sickle cell trait.
Two dated facts, nineteen years apart. We will not compress them into one tidy statistic — the two tables are not a like-for-like set and their criteria differ. But the direction of travel is not ambiguous, and consumer health copy has not caught up with it.
What this means if it might apply to you
Not all A1C tests are unreliable for people with a variant. The federal guide is explicit that someone getting false results from one type may need a different type — the fix is a different method, not a different patient. Elevated fetal hemoglobin does remain a live interference for several methods, which is one more reason the answer depends on which method your laboratory runs.
🩺 Physician Note: A common point of confusion is treating ancestry as the variable. It isn’t. Certain inherited variants are more common in some ancestries, and those variants affect some assay methods and not others. Current guidance is also clear that race and ethnicity are not good proxies for the underlying genetic differences and should not be used to decide how an A1C is interpreted.
✅ Patient Action: If a hemoglobin variant runs in your family, ask your primary care physician: “Does the method our lab uses handle hemoglobin variants — and is it worth checking whether I carry one?”
That question has a real answer, and it is easier to get than people assume.
How to find out if your own A1C result is solid
Three things — which is what fits into an appointment.
1. Ask which method your laboratory uses
This is not our advice. It is the recommendation of the standardization program and, independently, of the network coordinator interviewed by the federal institute: doctors or patients who want to know about the accuracy of a particular A1C method should first find out which method their laboratory is using.
Your clinician can ask the lab which test it runs, what interferences go with it, and whether any apply to you.
2. Compare like with like — the same lab, not just the same test
A certificate covers one instrument and one set of reagents. Two labs running different certified methods are both correct, and not interchangeable at the level of a 0.2 difference.
If you are tracking change over time, the comparison is only clean when both samples went to the same laboratory. A “rise” across two labs may be a change in method rather than a change in you.
🩺 Physician Note: Small differences and small changes in A1C need careful interpretation, because every assay method carries a degree of variability — and, as the coordinator of the national standardization network notes, clinicians often don’t realize it. This is not specific to A1C. It is true of every laboratory test.
3. Know when a repeat is reasonable — and when a different test is the answer
A repeat is part of how a diagnosis gets made, not a way to shop for a better number. When you have no symptoms and a test indicates diabetes or prediabetes, federal guidance is to repeat on a different day — using the A1C or another test — to confirm.
That decision belongs to your clinician, and a result you dislike is not by itself evidence it is wrong. When a blood result gets repeated, and why covers this across blood testing generally; if your figure landed in the 5.7–6.4 band, what doctors check next after a borderline result is the closer read.
✅ Patient Action: Take one question to your next appointment: “Which method does our lab run for A1C, and is there anything about my situation that could affect it?” Ask it before asking for a repeat.
One kind of result sits outside all of this — and it is the one people most often have in hand.
Finger-prick and at-home A1C: different rules apply
A fingerstick A1C at a clinic, a pharmacy screening and a kit bought online are not the same thing as the venous sample described above. The difference is not what most people assume.
Why a clinic fingerstick isn’t used to diagnose
Point-of-care results should not be used for diagnosis. When an A1C is used diagnostically, blood is drawn from a vein and sent to a laboratory running a certified method.
What “CLIA-waived” means for oversight
This gets reported unfairly in both directions. Many point-of-care devices are certified and, under ideal conditions, perform with a high degree of accuracy and precision — some match laboratory methods.
The problem is not the device. It is the setting.
Point-of-care methods are generally categorized as CLIA-waived. As the standardization network’s coordinator explains in an NIDDK interview on A1C and hemoglobin variants, waived settings carry much lower testing standards: the person running the test may have very little training, there is no site inspection, and there is no proficiency testing. The federal rules confirm that last point — proficiency-testing requirements do not apply to waived tests.
A diagnosis rests on the whole chain, not the device at the end of it.
Mail-in kits and fingerstick kits are not the same thing
These get routinely lumped together. A kit that ships your sample to a certified laboratory is a different product, under different rules, from a fingerstick device that produces a number on your kitchen table. How accurate at-home blood tests really are works that distinction through for cholesterol testing, and the reasoning transfers.
Why we don’t recommend an at-home A1C product
We do not recommend or link to any at-home A1C product, and we are not paid by any. Recommending one would contradict the sources cited above: federal guidance says point-of-care results are not for diagnosis, and the oversight gap is real.
The CDC’s guidance on following up a screening result gives the practical step — if you had your blood sugar tested at a health fair or pharmacy, follow up with your doctor, who may want to re-test you to make sure the results are accurate.
⚠️ Clinical Warning: Do not treat a fingerstick, pharmacy or at-home A1C as a diagnosis, in either direction. A reassuring home number does not rule out diabetes, and an alarming one does not establish it. Take it to a clinician and ask whether a venous sample is warranted before anything changes.
Common questions about A1C accuracy
1. How accurate is the A1C test?
Accurate enough to diagnose and monitor diabetes with. Before a method can be used, at least 36 of 40 patient samples must fall within 5% of a reference laboratory’s results, and the lab running it is graded on unknown samples three times a year. Small variation remains, which is why a diagnosis is confirmed with a second measurement.
2. Can an A1C test give a false reading?
Yes, though “false” usually means the number is accurate for your blood and misleading about your glucose. Conditions that shorten red blood cell survival, such as recent blood loss or hemolytic anemia, lower the result regardless of method, and kidney failure, liver disease or low iron can also distort it. Discuss any of these with your clinician before interpreting a result.
3. Why is my A1C different from my glucose meter reading?
They measure different things. A meter captures one moment; A1C reflects an average across roughly the past two to three months, so the two will rarely align neatly. A persistent mismatch is different — it is the recognized signal that something may be affecting the A1C measurement, and it is worth raising with your clinician rather than resolving yourself.
4. Do different labs give different A1C results?
They can differ slightly. A certification covers one named instrument running one named set of reagents, so two laboratories using different certified methods are both correct without being interchangeable at the level of a small difference. When you are tracking change over time, the comparison is cleanest when both samples went to the same laboratory.
5. What does NGSP-certified mean?
It means the method passed a documented forty-sample comparison against a reference laboratory and was issued a certificate of traceability to the reference method used in the Diabetes Control and Complications Trial. The certificate lasts one year and applies only to the specific reagents and instrument tested. It is why an A1C means the same thing across US laboratories.
6. Does sickle cell trait affect A1C results?
It depends entirely on the method. In the standardization program’s table of the twenty most-used methods, updated in June 2026, nearly every entry shows no clinically significant interference from sickle cell trait, and where a method is affected the answer is a different method rather than a different test. Ask your clinician which method your laboratory runs.
7. Can anemia make an A1C result wrong?
Yes, and the direction depends on the type. Iron-deficiency anemia is associated with a higher A1C and correcting the deficiency brings it back down, while anemias that shorten red blood cell survival push the result lower instead. Because the effect runs both ways, this is something to interpret with your clinician rather than adjust for yourself.
8. Is a finger-prick A1C as accurate as a lab test?
Many point-of-care devices are certified and, under ideal conditions, perform with high accuracy — some match laboratory methods. The gap is oversight, not hardware: waived settings involve no site inspection and no proficiency testing, while laboratories are graded three times a year. Federal guidance is that point-of-care results should not be used for diagnosis.
9. Are at-home A1C kits accurate?
We do not recommend any at-home A1C product, and it is worth separating two very different things: a kit that mails your sample to a certified laboratory, and a fingerstick device that produces a number at home. The second carries the oversight gap described above. Take any home or pharmacy result to your clinician before acting on it.
10. How do I ask which A1C method my lab uses?
Ask your clinician rather than the laboratory directly — they can query the lab about which test it runs, which interferences are associated with it, and whether any apply to you. Both the standardization program and the federal institute’s own expert recommend exactly this step for anyone questioning an A1C result’s accuracy.
11. Should I ask for a repeat if I think my A1C is wrong?
Raise it, but understand what a repeat is for. When you have no symptoms and a result indicates diabetes or prediabetes, repeating on a different day is how the diagnosis is confirmed — it is part of the process, not a second opinion. A result you dislike is not itself evidence it is wrong, and your clinician makes that call.
What to do with your A1C number now
Your A1C came from a method that had to prove itself against a reference laboratory, run by a lab graded on samples it cannot see coming. That system is not perfect, but its tolerances are tighter than those for most other numbers on the same page.
What it cannot see is you — your red blood cells, your iron, your kidneys, whether you carry a variant nobody has tested you for. That is what the conversation with your clinician is for, and it starts with one question about which method your laboratory runs.
If you want the full picture of what the number itself means, read your A1C result in full.
About this content
How this article was put together: researched from recognised health sources, drafted with the help of AI tools, and edited by hand, with sources linked throughout.
Sameer Patel is the founder and editor of My Medicine Advisor. He is not a doctor or medical professional — before starting this site he worked in banking,…
Medical disclaimer
The content on MyMedicineAdvisor is provided for general informational and educational purposes only and is not a substitute for professional medical advice, diagnosis, or treatment. Health information on this website should not be used to diagnose, treat, cure, or prevent any condition without guidance from a qualified healthcare professional. Always seek the advice of your doctor, physician, or another licensed healthcare provider with any questions you may have regarding a medical condition, symptoms, medications, or treatment decisions.
