HbA1c converter
Your report writes it in mmol/mol, your doctor speaks in percent and your meter reads mg/dL. Here are all three numbers, together with the range the true average glucose can actually fall in.
Most of Europe reports HbA1c in mmol/mol, while doctors, leaflets and websites still speak in percent. Pick the unit printed next to your number and you get all the others.
Conversion table, 5 to 13%
These are the official NGSP table values, rounded the way a laboratory prints them. The average glucose column comes from the ADAG equation and remains an estimate.
The two units of glycated haemoglobin
Glycated haemoglobin is the share of haemoglobin that has glucose attached to it, and it can be written two ways. The older one is the NGSP percentage, born in the DCCT and UKPDS trials of the 1990s and still used by doctors, leaflets and treatment targets. The newer one is the IFCC value in mmol/mol, millimoles of glycated haemoglobin per mole of total haemoglobin, which is what most European labs have printed on reports since around 2010. Both measure the same thing and are tied by the master equation agreed in 2007, HbA1c% = 0.09148 × mmol/mol + 2.152. That is where the pairs worth memorising come from: 42 mmol/mol = 6.0%, 48 = 6.5%, 53 = 7.0%, 64 = 8.0%. This tool inverts the equation by dividing by the coefficient rather than using the rounded 10.929 that circulates in handbooks, so it reproduces the official table row by row, including the rows that hinge on half a point.
From HbA1c to average glucose, the ADAG equation
Estimated average glucose (eAG) comes from the ADAG study published by David Nathan in Diabetes Care in 2008. The researchers followed 507 people, with type 1 diabetes, type 2 diabetes and no diabetes, for three months, measuring glucose with a continuous sensor plus seven fingersticks a day, and found a linear relationship with HbA1c: mg/dL = 28.7 × HbA1c% - 46.7. The mmol/L version follows by dividing by 18.0182, the molecular weight of glucose. The point is to translate a lab number into the number you read on a meter, so 7% becomes 154 mg/dL, that is 8.6 mmol/L.
Why the estimated average is not your average
The equation describes the average behaviour of a group, and around that average sits a wide spread that this tool always reports. In the ADAG study, 95% of the people with an HbA1c of 7% had a true average glucose between 123 and 185 mg/dL, a range more than 60 mg/dL wide. It depends on how long each person's red cells live and how readily their haemoglobin binds glucose. That is also why HbA1c says nothing about swings, hypos and variability: two people at the same 7% can live very different days, and only actual readings show that, not a formula.
Working backwards is shakier than working forwards
Starting from an average glucose to estimate HbA1c only works if that average really represents 24 hours. The average from a meter used mostly fasting or before meals sits lower than the true one, so the estimated HbA1c comes out understated. If instead the average comes from a continuous sensor, the right indicator is not the inverted ADAG but the Bergenstal GMI (2018), GMI% = 3.31 + 0.02392 × average in mg/dL, which the tool shows separately. The two routes diverge noticeably as averages rise: with an average of 200 mg/dL the inverted ADAG gives 8.6% while GMI gives 8.1%.
When HbA1c tells you less than it seems
The value reflects the last two or three months, but not evenly, since the most recent month carries roughly half the weight. And some conditions cost it its reliability, essentially all those that change how long red cells live. Haemolytic anaemia, recent bleeding, transfusions, advanced kidney or liver disease and pregnancy tend to push it down; untreated iron and B12 deficiency tend to push it up. Haemoglobin variants such as HbS, HbC and HbE can distort the measurement either way depending on the method the lab uses. In those situations HbA1c has to be read by someone who knows the clinical picture, or replaced with other tests.
Reference bands, not a diagnosis
The tool always shows every band and simply places your number, without passing judgement. Guidelines use 48 mmol/mol (6.5%) as the diagnostic criterion for diabetes, and an increased risk band starting at 39 mmol/mol (5.7%) for the American ADA and at 42 mmol/mol (6.0%) for the IEC/EASD criterion and the Italian AMD-SID standards, so the very same 5.8% is read two different ways depending on the document. A diagnosis is never made on a single value, since it needs confirmation on a second sample, and these thresholds do not apply to people already on treatment, in pregnancy or in childhood, where separate criteria and targets exist. This page is informative material, not medical advice, and it does not replace a professional opinion. Every calculation runs in your browser.