What Is a Normal Glucose Management Indicator (GMI) Percentage?

A glucose management indicator (GMI) is a percentage derived from continuous glucose monitor (CGM) data that estimates what your lab-measured HbA1c would be based on your average sensor glucose over a period of roughly two to four weeks. For most adults with diabetes, a GMI below 7% is the widely recommended target, mirroring the traditional HbA1c goal. But GMI and HbA1c are not interchangeable numbers, and the gap between them tells its own story about your metabolism that many people wearing a CGM never fully appreciate.

How GMI Is Calculated

GMI started life under a different name. Originally called “estimated A1c” (eA1C), the metric was renamed after the U.S. Food and Drug Administration determined that calling it an A1c estimate was misleading, since the two numbers often diverge in individual patients.¹ The current GMI formula takes your average CGM glucose and converts it to a percentage on the same scale as HbA1c. Your CGM app or reader does the math automatically using all available sensor data from the reporting period.

The calculation itself is straightforward: it is a linear equation that maps mean glucose in mg/dL to a percentage. A mean sensor glucose around 154 mg/dL translates to a GMI of roughly 7%. Lower average glucose means a lower GMI; higher average glucose means a higher one. The formula was built from clinical trials using the most accurate CGM systems available at the time, and it applies population-level averages to your individual data.¹ That distinction between population average and individual reality turns out to be the source of much confusion.

Target Ranges and What “Normal” Means in Practice

If you have diabetes and your care team has set a target HbA1c below 7%, a GMI below 7% is generally the corresponding CGM-derived goal. This aligns with the American Diabetes Association’s standard recommendation for most non-pregnant adults. Some people have tighter targets (below 6.5%), while others with complicated health situations or high risk of dangerous lows may aim for something more relaxed, around 7.5% or even 8%.

For people without diabetes who happen to be wearing a CGM, GMI values tend to cluster well below 7%. In fact, the GMI formula was designed and validated in people with diabetes, so applying it to a non-diabetic population can produce numbers that run slightly high relative to lab HbA1c. One study found that in people without diabetes, the mean GMI was about 0.6% higher than their actual laboratory HbA1c.² So if your lab A1c is 5.2%, your CGM might show a GMI around 5.8%. This does not mean your glucose control is worse than your blood test suggests; it means the formula was not calibrated for your glucose range.

Large-scale CGM mapping studies in thousands of healthy non-diabetic adults have generated reference percentiles for CGM-derived metrics across age and sex, providing a better sense of what “normal” sensor glucose patterns look like outside of diabetes.³ These resources can be more useful for non-diabetic CGM wearers than the GMI number itself, which is really built as a diabetes management tool.

Why GMI and HbA1c Almost Never Match

One of the most common surprises for CGM users is discovering that their GMI and their lab HbA1c are different numbers, sometimes by a lot. This discordance is not a bug or a sensor error; it is the expected result of the fact that HbA1c and GMI measure fundamentally different things. HbA1c reflects how much glucose has attached to your red blood cells over their lifespan (typically around 120 days). GMI reflects what your sensor saw over the past two to four weeks, run through a population-derived formula. Those are simply not the same measurement.

Research consistently shows that a close match between the two is the exception, not the rule. In one large real-world analysis, only about 11% of patients had an HbA1c and GMI within 0.1% of each other. Half had differences of 0.5% or more, and roughly one in five had a gap of 1% or greater.⁴ Another review found that over 80% of individuals had a discordance greater than 0.1%, and more than half differed by over 0.3%.⁵

Even when researchers compared GMI values derived from two different CGM systems worn simultaneously, moderate correlations with lab HbA1c were observed rather than tight agreement. One study in people with type 2 diabetes found correlation coefficients of about 0.71 for the Dexcom G4 and 0.68 for the Libre Pro.⁶ A correlation of 0.7 is decent for a biological measurement, but it means there is a lot of individual scatter. Your GMI and HbA1c can disagree substantially and neither number is necessarily “wrong.”

What Drives the Gap

Several factors push GMI and HbA1c apart, and they are worth understanding because they change what each number means for you personally.

Red blood cell lifespan: HbA1c depends on how long your red blood cells survive. If they turn over faster than average, less glucose accumulates on them, and your HbA1c reads lower than your actual average glucose would predict. Conditions like chronic kidney disease, certain blood disorders, and some medications can shorten red blood cell lifespan, dragging HbA1c down.⁷ In those cases, GMI may actually give a more honest picture of glucose exposure.
Kidney disease: Advanced chronic kidney disease (with significantly reduced kidney filtration) was specifically linked to greater discordance between HbA1c and GMI, likely because of its effects on red blood cell turnover and other metabolic factors.⁴
Race and ethnicity: Research suggests that GMI tends to run higher than HbA1c in white individuals but lower than HbA1c in Black and Chinese Asian individuals. This appears to reflect differences in glycation kinetics and other non-glycemic genetic factors rather than differences in actual glucose levels.⁸ The practical implication is that a Black patient whose GMI reads 7.0% might have an HbA1c that comes back at 7.5% or higher, which could affect clinical decisions if only one metric is considered.
Glycemic variability: How much your glucose swings up and down also influences the relationship. Higher variability weakens the correlation between GMI and lab HbA1c. In adults with type 1 diabetes, adjusting for glucose variability meaningfully changed the strength of the statistical relationship between the two.⁹
CGM accuracy and wear time: The sensor itself introduces some uncertainty. Different CGM brands read glucose at different intervals (every 5 minutes versus every 15 minutes), and how long you wear the sensor matters. One study found that at least 10 days of CGM data were needed for GMI to reliably agree with a full 14-day measurement in youth with type 1 diabetes.¹⁰

For clinicians and patients trying to reconcile a confusing mismatch, a published checklist of potential error sources for both GMI and HbA1c can help systematically work through what might be causing the gap.¹¹

GMI and Time in Range

Many CGM users find “time in range” (TIR) more intuitive than GMI. TIR tells you the percentage of time your glucose stays between target levels, typically 70 and 180 mg/dL. The two metrics are strongly correlated: in a study of children and adolescents with type 1 diabetes, GMI and TIR showed a high degree of linear correlation.¹² When GMI was around 7%, the predicted TIR was roughly 70–75%, depending on how variable glucose levels were.

The relationship gets less tight when glucose variability is high. At the same GMI, someone whose glucose fluctuates wildly will have a lower TIR than someone with steady levels, because those sharp spikes and drops eat into time spent in the target zone even if the average glucose is the same.¹² This is why most guidelines now encourage looking at both GMI and TIR together rather than relying on either in isolation. GMI gives you the average; TIR tells you how much of the day you are actually in a good zone; and the coefficient of variation (CV) captures how bumpy the ride is.

GMI in Pregnancy

Pregnancy changes glucose metabolism enough that standard targets and formulas need rethinking. Glucose targets in pregnancy are much tighter (a common CGM target range is 63–140 mg/dL rather than 70–180), and the physiological changes of each trimester affect both HbA1c and CGM readings differently.

In pregnancies complicated by type 1 diabetes, GMI and TIR correlated strongly across all three trimesters, with correlation coefficients around 0.9 for each. But the relationship between HbA1c and TIR was weaker during the first trimester, while GMI tracked TIR much more closely during that period.¹³ The implication is that early in pregnancy, GMI may better reflect glycemic control than HbA1c. The likely reason is that hemodilution and increased red blood cell turnover during the first trimester make HbA1c a less reliable marker. By the second and third trimesters, HbA1c and GMI aligned more closely with each other.

Children and Older Adults

The standard GMI formula was derived primarily from adult data, which raises reasonable questions about how well it works at the extremes of age. In young children (ages 0 to 6) with type 1 diabetes, a child-specific GMI formula was compared against the standard adult one. When at least 21 days of CGM data were used, there was no meaningful difference between the two formulas. With fewer days of data, the child-specific version tracked lab HbA1c more closely.¹⁴ The practical takeaway is that the standard formula works well enough in young children if you give it sufficient data, but short sensor-wear periods can produce less reliable GMI values in this age group.

Older adults present different challenges. International consensus groups have set CGM targets for older adults with diabetes, but these are based on limited evidence and rely heavily on expert opinion. The primary concern in this population is avoiding dangerously low glucose (hypoglycemia), which can cause falls, confusion, and cardiovascular events. Real-world data, mostly from relatively healthy older adults with type 1 diabetes, suggest that many can achieve standard CGM targets including TIR goals set for younger adults. However, meeting the target for time spent below range (less than 1%) remains harder. Reducing glucose variability appears to be a more effective strategy for avoiding lows than simply raising average glucose.¹⁵ For frail or medically complex older adults, the evidence base for CGM targets is especially thin, and GMI goals should be individualized rather than pulled from a standard chart.

Does the GMI-to-HbA1c Gap Predict Complications?

A question that researchers are actively working through is whether the ratio between GMI and HbA1c tells you anything useful about your risk of diabetes complications, beyond what each number says on its own. The logic sounds plausible: if your HbA1c is consistently higher than your GMI, your body is glycating hemoglobin more aggressively than average, and maybe that means your tissues are accumulating damage faster too.

The evidence so far is mixed and still evolving. One study in adults with type 1 diabetes found that the GMI-to-HbA1c ratio was not independently associated with diabetic retinopathy (eye disease) once HbA1c itself was accounted for. In other words, the gap did not carry extra predictive power beyond what HbA1c already provided.¹⁶ However, a separate longitudinal study explored an updated version of the GMI formula and found that this updated GMI had a somewhat stronger association with the development of diabetic retinopathy than HbA1c alone. The advantage was most apparent in a specific HbA1c range, roughly between 7.1% and 7.5%, where retinopathy risk escalated significantly according to the updated GMI but not as clearly according to HbA1c.¹⁷

These findings are not contradictory so much as they reflect different questions. The first study asked whether the ratio between the current GMI and HbA1c independently predicts eye disease, and the answer was no. The second study asked whether a refined version of GMI does a better job of identifying retinopathy risk than HbA1c in certain clinical gray zones, and the answer was yes. Neither study settles the question definitively, but together they suggest that future iterations of the GMI formula may eventually outperform HbA1c for predicting specific complications, particularly in borderline glucose ranges where clinical decisions are hardest.

The Updated GMI Formula

The GMI formula that most CGM systems currently use was published in 2018. Since then, researchers have proposed updated versions (sometimes called “uGMI”) that aim to better align with contemporary HbA1c measurements. The updated formula draws on newer and larger datasets and appears to reduce some of the systematic bias between GMI and lab HbA1c.¹⁸ Whether and when CGM manufacturers will adopt the updated formula remains to be seen. If you notice that your CGM software has changed its GMI calculation after a firmware or app update, this is likely the reason.

For now, the practical consequence is that any single GMI value you see on your CGM report is shaped by which formula version your device uses, how many days of data it draws from, and how your individual biology compares to the population the formula was built on. Treating GMI as a precise equivalent to HbA1c is the most common mistake people make with this metric. It is better understood as a rolling estimate of glucose exposure that complements, but does not replace, your lab work.

How Many Days of Sensor Data Are Enough

GMI reliability improves with more data, but there is a practical floor below which the number becomes unreliable. In one study of youth with type 1 diabetes, GMI values calculated from just a few days of data correlated reasonably well with a full 14-day measurement, but categorical agreement — meaning the GMI value would lead to the same clinical classification as the full period — only exceeded 90% when at least 10 days of data were available.¹⁰ In young children, using 21 days or more of CGM data produced GMI values that aligned well with lab HbA1c, while shorter windows were less reliable.¹⁴

Most CGM reports default to a 14-day window, which is a reasonable compromise between timeliness and reliability. If your sensor had significant gaps during that period due to warmup periods, lost connections, or early sensor failures, the GMI on your report may be less trustworthy. Some clinicians prefer to look at GMI calculated from a full 30 days when making treatment decisions. If you are reviewing your own data, pay attention to the “active sensor time” percentage — anything below about 70% of the reporting period starts to erode confidence in the calculated metrics.

Racial and Ethnic Differences in the Gap

The direction and magnitude of GMI-HbA1c discordance varies across racial and ethnic groups, and this has real implications for care equity. GMI is calculated purely from sensor glucose readings, so it does not carry any of the biological variation in hemoglobin glycation that can make HbA1c a less equitable metric. HbA1c has been shown to run higher in Black individuals at the same average glucose levels compared with white individuals, which has led to debates about whether HbA1c overestimates diabetes risk and complications in some populations.

GMI sidesteps this issue because it depends only on sensor glucose. When GMI reads lower than HbA1c for a particular person, that gap may indicate that their HbA1c is being influenced by non-glycemic biological factors. Some researchers have argued that in populations where HbA1c is known to carry racial bias, GMI could serve as a useful cross-check, though the current GMI formula has its own population-level assumptions baked in.⁸ Neither metric alone tells the complete story, and awareness of these systematic differences can help clinicians avoid over- or under-treating based on a single lab value.

1
PubMed Central. Glucose Management Indicator (GMI): A New Term for Estimating A1C From Continuous Glucose Monitoring
2
PubMed. Discordance Between Glucose Management Indicator and Glycated Hemoglobin in People Without Diabetes
3
Cell Metabolism. CGMap: A map of continuous glucose monitoring data in a large healthy non-diabetic population
4
PubMed Central. HbA1c and Glucose Management Indicator Discordance: A Real-World Analysis
5
PubMed. Understanding the clinical implications of differences between glucose management indicator and glycated haemoglobin
6
PubMed Central. Performance of the Glucose Management Indicator (GMI) in Type 2 Diabetes
7
PubMed Central. The influence of shorter red blood cell lifespan on the rate of HbA1c target achieved in type 2 diabetes patients with a HbA1c detection value lower than 7%
8
Pediatric Endocrinology Diabetes and Metabolism. Glucose management indicator – potential factors affecting differences in comparison with HbA1c and clinical significance of this phenomenon
9
PubMed Central. Impacts of glycemic variability on the relationship between glucose management indicator from iPro™2 and laboratory hemoglobin A1c in adult patients with type 1 diabetes mellitus
10
PubMed Central. Minimum Continuous Glucose Monitor Data Required to Assess Glycemic Control in Youth with Type 1 Diabetes
11
PubMed Central. Managing discordance between HbA1c and glucose management indicator
12
Diabetes. 1821-PUB: Relationship between Time in Range and Glucose Management Indicator in Adolescents and Children with Type 1 Diabetes Mellitus
13
PubMed Central. Relationship Between Time-in-Range, HbA1c, and the Glucose Management Indicator in Pregnancies Complicated by Type 1 Diabetes
14
PubMed Central. An Examination of the Glucose Management Indicator in Young Children with Type 1 Diabetes
15
PubMed Central. Glucose Targets Using Continuous Glucose Monitoring Metrics in Older Adults With Diabetes: Are We There Yet?
16
PubMed Central. The GMI/HbA1c ratio does not independently predict diabetic retinopathy in adults with Type 1 Diabetes
17
PubMed Central. Association of HbA1c and an updated glucose management indicator (uGMI) with incident diabetic retinopathy in adults with type 1 diabetes
18
PubMed Central. Updated glucose management indicator (GMI) better aligns with HbA1c than current GMI: implications for clinical practice and reporting