HbA1cGlycated haemoglobin · Haemoglobin A1c
A marker of your average blood sugar over roughly the past three months.
HbA1c (glycated haemoglobin) is the proportion of your haemoglobin, the oxygen-carrying protein inside red blood cells, that has glucose chemically stuck to it. The reaction is slow, one-way and unregulated: glucose in the blood attaches to haemoglobin at a rate that simply tracks how much glucose is there. Because red cells survive around three months before being replaced, the measurement acts as a running average of your blood sugar over roughly that window.
That averaging is the whole reason it displaced repeat glucose tests as the standard check. A finger-prick or fasting glucose captures one moment, and that moment is easily nudged by a poor night's sleep, a recent meal, an infection or the stress of being in a clinic. HbA1c cannot be gamed by fasting the night before, does not require you to fast at all, and reflects months rather than minutes, which is why it is used both to diagnose type 2 diabetes and to track how well established diabetes is being managed.
The UK reports it in IFCC units, mmol/mol, and has done since 2011; the older DCCT percentages still appear in American sources and in a lot of internet advice, which is a common source of confusion. Against WHO's 2011 criteria, as adopted by NICE, below 42 mmol/mol (6.0%) is the non-diabetic range, 42 to 47 mmol/mol (6.0 to 6.4%) is the band variously called prediabetes, non-diabetic hyperglycaemia or impaired glucose regulation, and 48 mmol/mol (6.5%) or above meets the threshold for type 2 diabetes. A diagnosis in someone without symptoms needs two separate readings, not one: a single result above a line is a reason to repeat the test, not a diagnosis.
The 42 to 47 band is the part worth understanding, because it is where the number is most useful and most often shrugged off. It is not a disease, and a good share of people in it never progress. It is a warning shot with a decade or more of runway, and the same measures that help there, losing excess weight, moving regularly, building and keeping muscle, cutting a chronic energy surplus, are the ones with the best evidence for stopping progression. In England, results in that band are what trigger referral to the NHS Diabetes Prevention Programme.
The averaging that makes HbA1c robust also makes it blind in a specific way. Two people with identical HbA1c can have very different days: one steady, one swinging between highs and lows that cancel out. That is why fasting glucose, and increasingly continuous glucose monitoring, add something HbA1c cannot: the shape of the variation rather than its average. Fasting insulin adds a third view again, since someone can hold a normal HbA1c only by producing a great deal of insulin to do it.
It is also genuinely unreliable in some circumstances, and this matters more than it is usually flagged. Anything that changes how long red cells live will distort the result: if cells are replaced faster than usual the reading is falsely low, and if they survive longer it is falsely high. That covers pregnancy, sickle cell trait and other haemoglobin variants, recent blood loss or transfusion, haemolysis, advanced kidney disease, and iron deficiency, which tends to push HbA1c up, so treating the iron deficiency can drop the number without any change in blood sugar at all. Where HbA1c is unreliable, glucose-based tests are used instead.
Sources
Where your own result would sit
Most people reading this page are holding a blood-test report with this marker on it, and the question they arrived with is whether their number is normal. A reference interval says where most results from a healthy population fall; a guideline threshold is where a clinical body draws a line for a decision. Neither is a target, and a result on the wrong side of one is not a diagnosis: the same number can be unremarkable in one person and worth looking into in another, which is why the clinician who ordered the test reads it against your history rather than against a list.
Everyone
- In range
- under 42 mmol/mol
- Borderline high
- 42 to 48 mmol/mol
- High
- 48 mmol/mol and over
Reported in IFCC units (mmol/mol), the UK standard since 2011, not the older DCCT percentage. HbA1c is unreliable and should not be read against this scale in pregnancy, in sickle cell trait and other haemoglobin variants, after recent blood loss or transfusion, and in advanced kidney disease.
Guideline decision points, not a laboratory range, from Diabetes UK. Read against that page on 2026-08-09. Your own laboratory may print a slightly different range.
On your own blood test?
HbA1c is one of the markers FitTools can read from an uploaded blood-test report. You confirm every value before anything is saved, your own laboratory’s printed range is used in place of the one above wherever your report carries one, and your results are explained together in one place.
Frequently asked questions
- What is a normal HbA1c in the UK?
- Below 42 mmol/mol (6.0%) is the non-diabetic range under WHO's 2011 criteria as adopted by NICE. From 42 to 47 mmol/mol (6.0–6.4%) is prediabetes, also called non-diabetic hyperglycaemia. From 48 mmol/mol (6.5%) upwards meets the threshold for type 2 diabetes, though in someone without symptoms that needs confirming with a second test.
- What is the difference between mmol/mol and the percentage?
- They are two scales for the same measurement. The UK switched to IFCC units (mmol/mol) in 2011; the older DCCT percentages are still standard in the United States and appear throughout online advice, which is why the same result can look like two different numbers. The key equivalences are 42 mmol/mol = 6.0% and 48 mmol/mol = 6.5%.
- Can HbA1c be wrong?
- Yes, and predictably so. Because it depends on red blood cells living about three months, anything that shortens their life makes it read falsely low, and anything that lengthens it makes it read falsely high. Pregnancy, sickle cell trait and other haemoglobin variants, recent blood loss or transfusion, haemolysis and advanced kidney disease all distort it. Iron deficiency tends to raise it, so correcting the iron can lower the number without your blood sugar having changed. Where HbA1c is unreliable, glucose tests are used instead.
- Do I need to fast for an HbA1c test?
- No. It measures an average built up over about three months, so what you ate that morning makes no meaningful difference. That is one of the practical advantages it has over fasting glucose.
- How quickly can HbA1c change?
- It lags real change by weeks, because it is an average over the lifespan of your red blood cells. Meaningful movement generally shows up over about three months, which is why repeat testing is usually spaced at roughly that interval. A change over a fortnight is more likely to be measurement variation than a real shift.
- Does prediabetes always turn into diabetes?
- No. A result in the 42–47 mmol/mol band means raised risk, not an inevitability, and many people never progress. It is the point at which the well-evidenced measures — losing excess weight, regular activity, keeping muscle, avoiding a chronic energy surplus — have the most runway to work. In England it is also the threshold for referral to the NHS Diabetes Prevention Programme.