Ask someone what their A1C is and most people can recite it the way they'd recite a credit score: a single figure that supposedly sums up how they're doing. Unlike a fasting glucose reading, which captures blood sugar at one moment on one morning, A1C is sold as something steadier. It is a blood test that shows the average level of blood glucose over the past two to three months, which is why physicians reach for it instead of a same-day snapshot when they want the fuller picture.
The mechanism is straightforward. Glucose in the bloodstream sticks to hemoglobin, the oxygen-carrying protein inside red blood cells, in a process called glycation. The more glucose circulating, the more of it attaches. Red blood cells live for about three months, so the percentage of glycated hemoglobin at any given moment reflects glucose exposure across the lifespan of the cells currently in circulation, rather than glucose from any one meal.
It is also, despite the "three-month average" label, weighted toward the present. The National Institute of Diabetes and Digestive and Kidney Diseases notes that although A1C results represent a long-term average, blood glucose levels within the past 30 days have a greater effect on the reading than levels from earlier months, because a portion of the circulating red blood cells are always relatively new. A large shift in control over the past four to six weeks moves the number more than most people expect.
What the number is used for
An A1C result is reported as a percentage, and the diagnostic bands are consistent across major reference sources: below 5.7 percent is normal, 5.7 to 6.4 percent signals prediabetes, and 6.5 percent or higher diagnoses diabetes, per MedlinePlus and the NIDDK. The test does not require fasting, since it reflects long-term exposure rather than a single meal. It screens for and diagnoses type 2 diabetes and prediabetes, and once a diagnosis exists, monitors treatment, typically with testing at least twice a year.
Some labs also report an estimated average glucose, or eAG: the A1C value converted into the same milligrams-per-deciliter units a home glucose meter uses. An A1C of 7 percent corresponds to an eAG of roughly 154 mg/dL. The eAG will not match any single glucose reading, since it, too, describes an average rather than a moment.
The assumption the test is built on
Here is the catch. The calculation depends on one unstated assumption: that red blood cells are living out their normal three-month lifespan, exposed to glucose the whole time, in the usual proportions. Anything that changes how long red blood cells survive changes the A1C reading even when a person's actual average glucose has not moved at all.
Conditions that shorten red blood cell survival, such as hemolytic anemia, recent blood loss, or a blood transfusion, tend to push A1C artificially low, since the cells are replaced faster and have had less time to accumulate glycation. Cleveland Clinic lists alcohol use disorder, hemorrhage, cirrhosis, pregnancy, and sickle cell anemia among the situations that can produce a falsely low result. Conditions that extend red blood cell life push the number artificially high instead. Iron-deficiency anemia is the most common example: the NIDDK flags it as a cause of falsely elevated A1C, alongside kidney failure and liver disease, and Cleveland Clinic adds vitamin B12 deficiency and thalassemia. None of this makes the test unreliable in general; it means the number reflects red blood cell biology as much as glucose, and the two can occasionally pull in different directions.
A second, separate blind spot: the hemoglobin itself
A different problem sits one layer deeper, in the structure of hemoglobin rather than its lifespan. Hemoglobin comes in several genetic forms beyond the common "hemoglobin A" most people carry. Hemoglobin C, D, E, and S traits are more common in people of West African, Central and South American, Caribbean, Southeast Asian, or Mediterranean descent, and a variant does not raise diabetes risk on its own. But it can interfere with certain A1C assay methods, producing a result that does not match a person's actual average glucose. The NIDDK is direct about this: people with a hemoglobin variant may need a different A1C test method, and providers may suspect interference specifically when A1C and a separate glucose test do not agree. Most people with a variant have no symptoms and no idea they carry it, which is why a mismatch between two tests prompts a second look rather than an assumption that one test is simply wrong.
An average can hide the shape of the day
Even when red blood cells and hemoglobin are entirely ordinary, A1C has a structural limitation that has nothing to do with error: it is an average, and averages erase shape. Cleveland Clinic makes the point plainly with an example: a person whose blood sugar swings sharply between very low and very high episodes throughout the day can land on the same A1C as a person whose blood sugar sits steadily near the middle of that range. Both might report a 7 percent A1C. Only one of them is living with the kind of volatility that matters for how they feel day to day, and for complication risk over time.
This is the gap that continuous glucose monitoring and the metric called time in range were built to fill. Time in range measures the percentage of a day spent inside a target glucose window, commonly 70 to 180 mg/dL, showing the swings an average necessarily smooths over. It is not a replacement for A1C, which still carries decades of outcome data behind it, but a complement: A1C answers "what was the average," and time in range starts to answer "how much did it move around to get there."
When two tests disagree
Because A1C and a same-day glucose test measure different things over different timeframes, they occasionally point in different directions: an A1C in the diabetes range with a normal fasting glucose, or the reverse. The NIDDK's guidance here is not to trust one test over the other by default, but to repeat testing, since a person with genuinely inconsistent results may be in an early stage of disease where glucose has not yet risen high enough to show up on every measurement. This is also why A1C alone is not used to diagnose type 1 diabetes, gestational diabetes, or cystic fibrosis-related diabetes, and why a diagnosis in someone without clear symptoms generally calls for a second confirming test on a different day rather than a single result.
Common questions
Do I need to fast before an A1C test?
No. The test reflects an average built over roughly three months rather than a single moment, so a recent meal does not meaningfully change the result, and blood can be drawn any time of day.
Can a normal A1C still miss diabetes?
Occasionally. Because different tests can catch the condition at slightly different points, a normal A1C sometimes appears alongside a glucose test result that suggests diabetes, or the reverse. This is more likely early in the disease, before glucose has risen high enough to register on every test, which is one reason providers often use more than one test rather than relying on A1C alone.
Why would my A1C and my glucose meter readings not match?
They answer different questions. A1C is a three-month average; a meter reading is a single moment. An estimated average glucose figure can translate an A1C percentage into meter-like units, but it will never match any individual reading, since it describes the average, not any one point on the curve.
Does a hemoglobin variant mean I'm at higher risk for diabetes?
No. Carrying a hemoglobin variant such as trait C, D, E, or S does not itself raise diabetes risk. It can, however, interfere with some A1C testing methods and produce a result that does not reflect true average glucose, which is why a mismatch between A1C and other glucose tests can prompt a provider to check for a variant or switch test methods.
Is A1C being replaced by continuous glucose monitoring?
Not replaced, but supplemented. A1C remains the primary test for diagnosis and management given the outcome data behind it, while time in range, drawn from continuous glucose monitoring, adds the day-to-day variability a single average cannot show.
