A1C to Average Glucose Converter
A1C converter. Enter NGSP %, IFCC mmol/mol, eAG mg/dL or mmol/L and get the other three, with the ADA bands and the conditions that make A1C unreliable.
A1C to Average Glucose Converter
Fill whichever box matches your report — the other three are worked out from it.
Please tick the acknowledgement above before calculating.
- Anaemia — iron-deficiency, haemolytic, or following blood loss
- Haemoglobin variants such as HbS, HbC, HbE, or thalassaemia
- Pregnancy
- Chronic kidney disease, especially on dialysis or erythropoietin
- A red-cell transfusion in the last three months
- Any condition that alters red-cell lifespan
The equations, in full
Estimated average glucose: eAG (mg/dL) = 28.7 × A1C − 46.7, from the
ADAG study. The mmol/L figure is derived from that result rather than from a
second equation, so the two cannot drift apart.
Unit conversion: NGSP % = 0.09148 × IFCC + 2.152 and
IFCC = 10.93 × NGSP − 23.50, both taken verbatim from the NGSP
master equation. They are not exact inverses of one another, and we have kept
them as published rather than deriving one from the other.
Nothing you enter is stored or transmitted. The conversion runs in your browser.
How to Use the A1C Converter
Find your unit
A US lab reports a percentage. A UK, EU or most Asian lab reports mmol/mol. Use whichever box matches your report.
Type it in
The other three boxes fill in as you type. All four are peers — none is more "real" than the others.
Read the caveats first
They sit above the result on purpose. If any apply to you, the A1C itself may not mean what it appears to.
Take it to a clinician
A band is not a diagnosis, and a target is individual. Neither is something a converter can decide.
One Result, Four Ways of Writing It
An A1C result is a single measurement that the world has agreed to write down in four different ways, and the differences are large enough to be alarming if you do not know what you are looking at. A result of 7% is the same result as 53 mmol/mol. Those two numbers share no digits and differ by a factor of seven, and a reader who has moved between countries — or who is reading a study published somewhere other than home — has every reason to be briefly confused.
The percentage is the older convention, from the American standardisation programme that first made A1C results comparable between laboratories. The mmol/mol figure is the internationally standardised unit: millimoles of glycated haemoglobin per mole of total haemoglobin. It is a genuine ratio rather than a percentage, which is why the numbers look so different. The UK switched to reporting it in 2009, much of Europe and Asia followed, and the United States did not — so both live on, and cross-border readers need both.
Estimated average glucose is a third way of expressing the same underlying reality, translated into the units people actually use day to day. It comes from a 2008 study that put continuous glucose monitors on 507 people — some with type 1 diabetes, some with type 2, some with neither — and tracked them for about three months to see how measured average glucose related to A1C. The result was a simple straight line, and it is the line this converter uses. Because glucose itself is reported in mg/dL in the United States and mmol/L nearly everywhere else, eAG immediately becomes two more units, and we arrive at four.
What the number does and does not tell you
A1C looks back roughly eight to twelve weeks, weighted towards the most recent of them. That makes it useful in a way a fingerstick is not: it cannot be gamed by one careful week, and it does not care what you ate this morning. But an average conceals its own spread. Two people with the same A1C can have very different days — one steady, one swinging between highs and lows that cancel out on paper. This is exactly why continuous monitoring reports time-in-range alongside an average, and why a good A1C does not always mean a comfortable one.
When the number is not to be trusted at all
The A1C-to-glucose relationship assumes ordinary red cells living an ordinary length of time. When that assumption breaks, so does the result. Anaemia of most kinds, haemoglobin variants such as HbS, HbC, HbE and thalassaemia, pregnancy, chronic kidney disease — especially on dialysis or erythropoietin — and any transfusion in the previous three months can all push A1C away from the truth, sometimes upward and sometimes downward.
That list deserves more prominence than it usually gets, and on this page it sits above the result rather than beneath it. Haemoglobin variants are not a rare curiosity: they are common across Southeast Asia, South Asia, Africa and the Mediterranean, which between them account for a very large share of the world. A reader in Kuala Lumpur or Jakarta is meaningfully more likely to be affected than the average American reader for whom most A1C explainers are written. An A1C that is simply wrong is more dangerous than no A1C at all, because it arrives wearing the authority of a laboratory result.
A band is not a diagnosis
The ADA's thresholds — normal below 5.7%, prediabetes 5.7% to 6.4%, diabetes 6.5% and above — are genuinely useful for orientation, and this tool shows you which band a value falls in. What it will not do is tell you that you have diabetes. The ADA itself normally requires a second confirmatory test on a separate day, and that means two real laboratory results, interpreted alongside symptoms and everything else known about you. It emphatically does not mean one number typed into a web form. Where a value lands in the diabetes range, this page says so plainly rather than letting the colour of a box imply a conclusion.
10 Facts About A1C and Average Glucose
A1C reflects roughly the last 8–12 weeks — weighted towards the most recent ones.
The eAG equation comes from 507 people wearing continuous monitors for three months.
7% A1C is 53 mmol/mol. The two scales look nothing alike, which is the whole problem.
IFCC units are mmol of HbA1c per mol of Hb — a ratio, not a percentage.
The ADA threshold for diabetes is 6.5%, which is 48 mmol/mol.
Prediabetes is 5.7%–6.4% — a risk state, not a diagnosis.
A diagnosis normally needs a second test on a different day.
Anaemia and haemoglobin variants can make A1C read falsely high or falsely low.
In pregnancy, red-cell turnover shifts and A1C understates true glycaemia.
eAG is an average. Two people with identical A1C can have very different swings.
Frequently Asked Questions
- It measures the proportion of your haemoglobin that has glucose stuck to it. Glucose attaches to haemoglobin slowly and more or less irreversibly, so the percentage that has been modified reflects how much glucose has been circulating over the lifespan of your red cells. That gives roughly a two-to-three month view, weighted towards the most recent weeks — the last month contributes considerably more than the month before it. It is not a snapshot and it is not an average of your fingerstick readings; it is a different measurement that happens to correlate strongly with your average glucose.
- With a regression published in 2008 by the A1c-Derived Average Glucose study group: eAG in mg/dL equals 28.7 times your A1C minus 46.7. The study followed 507 people with type 1 diabetes, type 2 diabetes, and no diabetes, who wore continuous glucose monitors and did frequent fingersticks over about three months, so the relationship is derived from measured glucose rather than assumed. This calculator uses that equation exactly as the American Diabetes Association publishes it, and derives the mmol/L figure from the mg/dL one so the two can never drift apart.
- History and geography. The percentage figure — the NGSP unit — came out of the American standardisation programme and is what the United States still reports. The IFCC unit, mmol of HbA1c per mol of haemoglobin, is the internationally standardised measurement and is what the UK, much of Europe and a good deal of Asia now use on lab reports. Estimated average glucose then exists in two flavours because glucose itself is reported in mg/dL in the US and mmol/L nearly everywhere else. So a single result can be written four ways, and a reader in Singapore holding a mmol/mol figure gets little help from a converter that only offers the American pair.
- They are linked by the NGSP master equation. Going one way, NGSP % equals 0.09148 times the IFCC value plus 2.152; going the other, IFCC equals 10.93 times the percentage minus 23.50. This calculator uses both of those exactly as the NGSP publishes them rather than algebraically inverting one to get the other. That is a deliberate choice: the two published forms are not perfect inverses of each other — 10.93 against 1 divided by 0.09148, which is 10.9313 — and matching the authority matters more than internal tidiness. The difference is far too small to affect a clinical decision, but it is real, and pretending otherwise would be the sort of quiet inaccuracy this site tries to avoid.
- The American Diabetes Association places normal below 5.7%, prediabetes from 5.7% to 6.4%, and diabetes at 6.5% and above. The edges belong to the higher band, so exactly 5.7% is prediabetes and exactly 6.5% is in the diabetes range. Prediabetes is worth understanding properly: it is a risk state rather than a diagnosis of disease, and it is the range in which lifestyle change has the largest documented effect on whether someone goes on to develop type 2 diabetes. It is a signal to act, not a verdict.
- Not on its own. The ADA normally requires a second, confirmatory test on a separate day before diagnosing diabetes — the exception being when blood glucose is very high or classic symptoms are present alongside one positive test. And that is about two real laboratory results, not one number typed into a web page. This tool converts units and shows you which band the value falls in; it has no idea whether your sample was handled properly, whether your assay is affected by a haemoglobin variant, or what the rest of your clinical picture looks like. Take the number to a doctor rather than to a conclusion.
- Whenever something interferes with red cells or with the assay. Anaemia of most kinds, haemoglobin variants such as HbS, HbC, HbE and thalassaemia, pregnancy, chronic kidney disease — particularly on dialysis or erythropoietin — a transfusion within the last three months, and anything else that shortens or lengthens red-cell lifespan can all push A1C away from the truth, in either direction. This matters more than it sounds: haemoglobin variants are common in parts of Southeast Asia, the Mediterranean, Africa and South Asia, so "A1C is unreliable in variants" is not a footnote for rare cases. If any of these apply to you, the conversion this page performs is still arithmetically correct, but the A1C going into it may not represent your actual glucose.
- Loosely, and with care. Your meter reads a moment; eAG estimates an average across months. They can differ substantially for perfectly innocent reasons — most people test more often before meals than after them, which biases the fingerstick average downwards relative to the true 24-hour mean that A1C reflects. A meaningful comparison needs readings spread across the day and night, which is really what continuous monitoring is for. If your eAG and your meter average disagree by a lot and the sampling explanation does not fit, that is worth raising rather than explaining away.
- No, and this is a genuine clinical trade-off rather than a hedge. Driving A1C down aggressively increases the risk of hypoglycaemia, which carries its own serious harms, and the balance shifts with age, how long someone has had diabetes, other conditions, and which medications they take. Guidelines reflect this by setting individualised targets rather than one number for everyone — a fit person recently diagnosed and an older person with cardiovascular disease are not aiming at the same figure. What your target should be is a conversation with your own clinician, and it is not something a converter can tell you.
- Because 28.7 times 6 minus 46.7 is exactly 125.5, and the ADA rounds that to 126 in its published table. The catch is that computers do not store 125.5 exactly here — in standard floating-point arithmetic the result comes out as 125.49999999999997, which naive rounding turns into 125. We found this while checking our output against the ADA table rather than against ourselves, and corrected the rounding so the two agree. If another converter shows you 125, that is almost certainly what happened to it.
- No. The conversion runs entirely in your browser. Nothing you type is transmitted to us, stored on your device, or logged. There is no account, and there is no record of your result anywhere after you close the tab.
Related News
You may be interested in these recent stories from our newsroom.
No related news yet for this tool. Our editorial team publishes new pieces every week.
Browse all news →Pick up where you left off
Stored only in this browser — never sent to our servers.