A difference between them is normal rather than a malfunction. This guide covers how each number is produced, where a gap comes from, what a gap can and cannot tell you, and what to bring to a clinician.
A1c measures the share of hemoglobin inside your red blood cells that has glucose chemically attached to it. Hemoglobin A1C is run on a blood sample, needs no fasting, and reflects glucose exposure over roughly the previous two to three months, weighted more heavily toward the most recent weeks 1. Labs report it as a percentage of total hemoglobin. It is used to screen for glucose problems and to follow glucose patterns over time, and a single value is an average, not a diagnosis on its own. The biomarker page linked above owns the detail on how A1c is tested, what typical and optimal ranges look like, and how to read one result. The rest of this guide stays on the comparison with GMI.
GMI, the glucose management indicator, is a single percentage calculated from the average glucose a continuous glucose monitor has recorded over a defined stretch of days. Nothing extra is measured in your blood to produce it. The sensor samples glucose in interstitial fluid every few minutes, the software averages those readings across the chosen window, and the average is converted onto the familiar A1c percentage scale using a published equation.
The equation that introduced the term is GMI (%) = 3.31 + 0.02392 x mean glucose in mg/dL, derived from paired CGM and laboratory A1C data 2. The label GMI was proposed in place of the older wording 'estimated A1C', because that phrasing encouraged people to read a calculated value and a laboratory measurement as the same quantity 2. The equation was built around continuous glucose data collected over a 14 day window 2.
The practical consequence of that design is worth sitting with. Mean sensor glucose is the only input. Two people with identical average glucose over the same number of days will be given an identical GMI, regardless of how long their red blood cells live, how readily their hemoglobin picks up glucose, or what their laboratory A1c would actually show. GMI is a restatement of mean glucose in A1c units, not a second measurement of glycation.
That is also its strength. Because it is anchored to sensor readings, GMI updates as soon as there is enough data, it can be recalculated for any window, and it is not influenced by anything happening inside red blood cells.
The two numbers differ in what is measured, where the measurement happens, how far back it looks, and what can distort it. The table below lays the differences side by side.
| Feature | GMI | A1c |
|---|---|---|
| What is measured | Glucose in interstitial fluid, sampled by a sensor every few minutes | Glucose chemically bound to hemoglobin inside red blood cells |
| How the number is produced | An equation applied to mean sensor glucose | A laboratory assay run on a blood sample |
| Time window | Only the days of CGM data included in the report | Roughly the previous two to three months, weighted to recent weeks |
| Reported as | A percentage, on the same scale as A1c | A percentage of total hemoglobin |
| Equipment needed | A continuous glucose monitor with enough wear time | A blood sample, no device required |
| Influenced by red cell lifespan | No | Yes |
| Influenced by sensor accuracy and data gaps | Yes | No |
| Used in diagnosis | No | Yes, by a clinician, alongside other information |
Read that table once more and the headline point becomes obvious. These are not two attempts at the same measurement. They are two different measurements that happen to be printed in the same units, which is exactly why people expect them to agree and are surprised when they do not.
The same person gets two different numbers because each one is produced from different biology over a different period, and because the conversion between them is built on a population average rather than on your individual physiology.
Four separate mechanisms contribute.
Different raw material. A1c depends on how much glucose has attached to hemoglobin, which depends on both the glucose level and how long each red blood cell stays in circulation. GMI depends only on the glucose level the sensor saw. A person whose red cells survive longer than average accumulates more glycated hemoglobin at any given average glucose, which nudges the laboratory A1c upward relative to the calculated GMI.
Different time windows. A CGM report covering the last 14 days describes the last 14 days. A laboratory A1c drawn on the same morning describes a far longer look back. If the last fortnight was unusual, quieter travel, a change in routine, illness, or a shift in eating patterns, the two numbers describe two genuinely different stretches of your life.
Population averages applied to individuals. The conversion from mean glucose to an A1c-scale percentage was fitted across a group of people, and research suggests that reading the result as a direct substitute for a laboratory A1C can mislead at the individual level 3. Studies indicate that in people with type 2 diabetes, GMI and measured HbA1c can diverge enough at the individual level to change how a report is interpreted 4.
Measurement conditions on both sides. Sensor accuracy, wear time and the number of days included all shape the mean glucose that feeds the equation, and studies indicate that the GMI value shifts depending on the sampling duration used to calculate it 5. On the laboratory side, assay method and anything that alters red blood cell turnover influence the measured value 1.
Research suggests that GMI and a laboratory A1C from the same person frequently fail to match, which is why the term was deliberately separated from the phrase 'estimated A1C' in the first place 2.
A gap is an observation about how your glucose is being summarised, not a verdict about your health. The useful question is which direction the gap runs, and whether it persists across repeat testing.
When the laboratory A1c runs higher than GMI. The observation is that your hemoglobin carries more glucose than your average sensor glucose alone would predict. That may reflect longer red blood cell survival, a greater tendency for your hemoglobin to glycate at a given glucose level, a CGM window that missed a higher stretch, or glucose excursions during periods when the sensor was not being worn. It is worth discussing with a clinician, particularly if the pattern repeats over several months.
When the laboratory A1c runs lower than GMI. The observation is that your hemoglobin carries less glucose than the sensor average would predict. Shorter red blood cell survival is associated with a lower A1c for the same average glucose, and conditions that speed red cell turnover or alter hemoglobin are known reasons A1c under-represents glucose exposure 1. Again, the next step is a conversation, not a conclusion.
When the gap is small. Small differences are unremarkable. The two numbers were never designed to land on the same decimal.
One more point deserves care, because the gap itself is sometimes promoted as a risk signal. An analysis in adults with type 1 diabetes found that the GMI to HbA1c ratio did not independently predict diabetic retinopathy 6. The honest reading is that a gap tells you something about how your two numbers are generated, and that it is not, by itself, a measure of long term risk.
A1c has established laboratory reference ranges. GMI does not, because it is a calculated output of a device rather than a measured analyte.
| Marker | Typical lab reference range | FitScript optimal range |
|---|---|---|
| Hemoglobin A1C | 4 to 5.7 % | 4.6 to 5.3 % |
| GMI | No laboratory reference range. Calculated from CGM mean glucose and reported on the A1c scale | Not applicable. GMI is a calculated CGM output, not a measured lab value |
These are typical lab reference ranges, and they vary by lab, age and sex, so read your own report against the range printed on it. The optimal range shown is FitScript's optimization target, not a diagnostic cut-off, and it is narrower than the standard range by design.
Because GMI has no reference range of its own, it is read against targets a clinician sets with you, and alongside the other things a CGM report shows that a single percentage cannot, including time in range, the shape of daily glucose curves, and how often glucose drops low. A GMI of a given value built on steady, flat glucose and the same GMI built on large swings describe very different days.
Yes. The window you choose changes the number. Studies indicate that GMI calculated over different sampling durations yields different values, so two reports generated from the same sensor on different windows can disagree 5.
The arithmetic behind this is simple. A mean calculated over 7 days is more easily pulled by a single unusual day than a mean calculated over 30. Sensor wear matters in the same way. If the sensor was off for a third of the window, the average describes only the hours it was on, and those hours may not be representative, particularly if the gaps cluster around meals, travel or sleep.
This is also why comparing a GMI from one month to a GMI from another month is only fair when the windows are of similar length and the wear time is similar. Before concluding that a number moved, check that the two reports were built the same way. The date range and the percentage of time the sensor was active are usually printed at the top of the report, and both belong in any comparison.
Several things unrelated to your average glucose can shift a laboratory A1c, because the test depends on red blood cells as much as it depends on sugar.
Conditions that shorten or lengthen red cell survival change how much time hemoglobin has to accumulate glucose. Anemias, hemoglobin variants, recent blood loss, recent transfusion, pregnancy and chronic kidney disease are among the situations in which A1c is a less reliable mirror of average glucose, and clinical guidance on glucose monitoring treats these as reasons to interpret the value cautiously or to rely more on direct glucose measurement 1.
Assay interference is a separate issue. Certain hemoglobin variants can affect some measurement methods more than others, which is one reason a surprising A1c is sometimes worth repeating, occasionally by a different method, at a clinician's direction.
None of this makes A1c a poor test. It makes A1c a test with known conditions attached, which is precisely the kind of thing worth knowing before you compare it to a number generated by a sensor on your arm. If your A1c and your blood count results are both available, they are more informative read together than apart.
GMI inherits every limitation of the data feeding it, and adds the limitation of the equation itself.
Sensor wear and data gaps. An average built on partial data represents only the hours captured. Low wear time is the single most common reason a GMI misrepresents a period.
Sensor accuracy and drift. Sensors measure interstitial fluid, which lags blood glucose during rapid changes, and individual sensors vary in how closely they track reference values. A consistent sensor bias shifts the mean, and therefore the GMI, in one direction.
Window length. Covered above, and worth repeating because it is so easily missed, since studies indicate the calculated value depends on the sampling duration used 5.
The equation's population basis. The conversion was fitted to a group, and research suggests the resulting number should be read as a summary of CGM glucose rather than as a replacement for a laboratory measurement 3.
The practical takeaway is that GMI is excellent for tracking yourself against yourself over comparable windows, and weaker as an absolute statement about what a blood test would show.
Estimated average glucose runs the conversion in the opposite direction. It takes a measured laboratory A1c and expresses it as the average glucose that value corresponds to, which makes a percentage easier to picture in mg/dL or mmol/L.
So the three numbers line up like this. Mean sensor glucose is measured, and GMI converts it into A1c units. A1c is measured, and estimated average glucose converts it into glucose units. Each conversion carries the same caveat in both directions, because both rest on an average relationship across many people rather than on your individual glycation.
That is why a sensor average and an estimated average glucose derived from a lab A1c can land in different places for the same person during the same month. Neither figure is wrong. They were produced from different measurements, and converted using relationships that hold better across groups than within one body.
Laboratory A1c remains the measured value used in diagnosis and in long term follow up, while CGM metrics describe day to day glucose behaviour in detail that a single percentage cannot 1. The two serve different jobs.
A1c is standardised, repeatable across labs, and requires no device, which makes it the anchor for formal assessment and for comparing a person with themselves over years. It cannot show when glucose rises, how far it swings, or how often it drops low.
CGM data shows exactly those things, and GMI is a convenient headline for the glucose average underneath. It is best used to track direction over comparable windows, to sanity check whether a period went as expected, and to give context to a lab result rather than to replace it. Research suggests caution in reading GMI as a stand-in for a laboratory A1C at the individual level 3, and studies indicate the same caution applies in people with type 2 diabetes 4.
Any decision about what to do next, including whether any therapy is appropriate, belongs to a licensed provider who has seen your full picture.
Glucose markers answer more when they are read as a group, because each one has a blind spot the others can cover.
FitScript's membership is built on this kind of grouped testing, with 150+ biomarkers tested per year and interpretation by Atlas, FitScript's AI health advisor, so a single out of pattern glucose number is read next to the blood count, iron status and insulin markers rather than on its own.
Start by checking that you are comparing like with like, then take both reports to a clinician rather than deciding between them yourself.
Three checks are worth doing first. Confirm the date range on the CGM report and how much of that window the sensor was actually active. Confirm when the blood sample was drawn, since a lab A1c reaches back months while a CGM report may cover a fortnight. Confirm that the GMI windows you are comparing across months are the same length.
If the gap survives those checks and repeats on a second round of testing, it is a genuine finding about your physiology, and it is useful. It may mean your laboratory A1c systematically reads higher or lower than your sensor average, in which case a clinician can decide which number to follow for you, and whether anything about your red blood cells deserves a closer look.
What it does not mean is that your device is faulty or that your lab made an error. Both can be right at once.
It is worth booking a conversation when either number sits outside the standard range, when the gap between them is large or persistent, or when symptoms are present.
Specifically, bring it up when an A1c sits above the typical reference range, when a CGM report shows frequent or unrecognised low glucose, when either number has moved substantially without an explanation, or when you have symptoms such as unusual thirst, frequent urination, blurred vision, unexplained fatigue or unexplained weight change. A suboptimal value with symptoms attached is worth discussing with a clinician promptly, and no result should be ignored because the other number looked better.
Discuss your results, and any next step, with a clinician or licensed provider who can see your full history, since only a licensed provider can decide whether any therapy is appropriate for you. This page is educational and is not medical advice.
For the full marker page, including how the test is run and how to read a single result, see Hemoglobin A1C. To convert a percentage into glucose units, use the A1c to average glucose calculator, and for what that converted figure represents, see estimated average glucose. If your result sits in the upper part of the standard range, is hemoglobin a1c 5 7 bad covers how that value is read, and how to lower your hemoglobin a1c covers the levers that influence it.
No. GMI is a percentage calculated from the average glucose your continuous glucose monitor recorded over a chosen window. A1c is a laboratory measurement of how much glucose is attached to hemoglobin in your red blood cells. They are printed in the same units, which makes them look interchangeable, but they come from different measurements covering different periods.
Neither replaces the other, because they answer different questions. A1c is a standardised laboratory measurement and is the number used in formal assessment. GMI is a summary of what your sensor actually saw over a specific set of days. For tracking yourself against yourself over comparable windows, GMI is useful. For a measured value a clinician can compare across years and labs, A1c is the anchor.
It can mean your hemoglobin carries less glucose than your sensor average would suggest, which is sometimes related to how long your red blood cells survive. It can also mean the CGM window covered a higher stretch than the months your A1c reflects. Check the dates on both reports and the sensor wear percentage first, then take both to a clinician if the pattern repeats.
Your hemoglobin may be carrying more glucose than your sensor average predicts, which can relate to red blood cell lifespan or to glucose excursions during hours the sensor was not worn. A CGM report also looks back a fortnight or so, while an A1c looks back months. A clinician can decide whether the difference needs any follow up.
More days and higher sensor wear give a steadier number, since the value is just an average dressed up as a percentage. The window length used to produce the number changes it, so compare reports built on windows of similar length with similar wear time. Your report should show both the date range and the percentage of time the sensor was active.
No. GMI is a calculated device output, not a diagnostic test, and no single number diagnoses anything by itself. Diagnosis is made by a clinician using measured laboratory results together with your symptoms and history. Bring your CGM reports along, because they add detail a lab value cannot show, but the diagnostic decision rests on laboratory testing.
Usually not. A gap is expected, because the two numbers are produced from different biology over different time windows and converted using a relationship fitted across many people. Before suspecting the device, check sensor wear time, data gaps and whether the two reports cover comparable periods. If a sensor seems consistently off against fingerstick checks, raise it with the manufacturer and your clinician.
Yes, unless your clinician advises otherwise. A laboratory A1c is a measured value that can be compared across labs and across years, and it is the number used in formal assessment. Your CGM shows what your glucose does hour to hour. Having both gives a clinician far more to work with than either one alone.
GMI has no laboratory reference range of its own, because it is a calculated output rather than a measured analyte. It is read on the A1c scale and against targets a clinician sets with you, alongside the rest of the CGM report such as time in range and how often glucose drops low. A single percentage without that context tells you less than it appears to.