Estimated Average Glucose translates HbA1c into an equivalent daily average blood glucose reading. It's mathematically derived from A1C, so the two markers always agree.
eAG gives HbA1c an intuitive scale. Instead of '5.4%' it reads as '108 mg/dL daily average.' Useful for patient communication and for comparing continuous glucose monitoring output against long-term glycemic control.
Optimal Estimated Average Glucose: 85-105 mg/dL.
Typically re-tested every 12 weeks.
It appears next to A1c on many lab reports because most people find mg/dL easier to picture than a percentage.
Estimated average glucose is a calculated value that translates a hemoglobin A1c percentage into the average blood glucose concentration, in mg/dL, that would typically produce it.
The biology behind it is simple. Glucose circulating in your blood attaches to hemoglobin inside red blood cells in a slow, non-enzymatic reaction called glycation. The higher the surrounding glucose and the longer the exposure, the larger the glycated fraction. Because a red blood cell survives for roughly 120 days, the A1c percentage behaves like a weighted average of glucose exposure over the preceding two to three months, with the most recent weeks counting more heavily. Labs regard A1c as a stable, non-fasting summary of that exposure rather than a snapshot[1].
eAG is not a separate blood test. It is arithmetic applied to the A1c already measured, using a conversion derived by comparing A1c values against continuous glucose monitoring and frequent fingerstick readings. The formula is eAG (mg/dL) = 28.7 × A1c (%) minus 46.7. Because the number is derived, it carries every strength and every limitation of the A1c underneath it[2].
One point confuses people often. eAG is not the average of your morning fasting readings. It folds in overnight glucose, post-meal rises, and everything in between, so it usually sits higher than a string of fasting values would suggest.
Related: When to Check Blood Glucose? (Fasting vs After Meals)
It matters because months of glucose exposure, not single readings, track most closely with the slow changes seen in blood vessels, nerves, kidneys and eyes.
Sustained higher glucose exposure is associated with both microvascular and macrovascular complications, and long-term glycaemic summaries have been the main outcome measure in decades of diabetes research[2]. Large trials also taught a more nuanced lesson.[3] A review of the Veterans Affairs Diabetes Trial concluded that pushing glycaemic targets lower produced benefits that were real but modest for large-vessel outcomes, and came with a meaningful rise in low-glucose episodes, which is why targets are individualised by a clinician rather than applied uniformly[3].
For people earlier in the metabolic story, the value of eAG is mostly directional. Risk related to glucose appears to rise across a continuum rather than switching on at a single threshold, so watching the number drift year to year is often more informative than any one result. Research in people with type 2 diabetes not on insulin suggests that structured feedback on glucose levels is associated with improved glycaemic control and better cardiovascular risk factors[4].
There is also something eAG cannot show. Two people with the same average can have very different glucose patterns, one steady and one swinging. That is why attention has shifted toward how much time glucose spends inside a target band, including work on tighter target ranges using real-world continuous monitoring data[5], and why continuous sensors developed alongside closed-loop systems have become central to glucose research[6].
Most labs report estimated average glucose against a broad reference range, while optimization targets sit considerably tighter.
| Marker | Typical lab reference range | FitScript optimal range |
|---|---|---|
| Estimated average glucose | 82 to 154 mg/dL | 85 to 105 mg/dL |
eAG has no reference intervals by sex and age of its own, so these figures are HbA1c, the lab value eAG is calculated from. They come from healthy adults in two German population studies (SHIP-0 and SHIP-Trend, 8,665 participants combined), where HbA1c rose with age.[13]
| Sex | Age | 2.5th to 97.5th percentile (HbA1c) (% HbA1c) |
|---|---|---|
| Men | 20 to 39 | 4.1 to 6.1 |
| Men | 40 to 59 | 4.0 to 6.2 |
| Men | 60 and older | 4.4 to 6.6 |
| Women | 20 to 39 | 4.0 to 5.8 |
| Women | 40 to 59 | 4.1 to 6.1 |
| Women | 60 and older | 4.4 to 6.5 |
These are HbA1c reference intervals from healthy people, not eAG values and not the cut-offs used to diagnose diabetes. Your own lab's printed range is the one to read your result against.
Reference ranges vary by lab, assay method, age and sex, so always read your result against the range printed on your own report. FitScript's optimal range is an optimization target, not a diagnostic cut-off, and it is narrower than the standard range by design.
Because eAG is calculated, it maps directly onto A1c. The table below shows the usual conversions.
| A1c (%) | Estimated average glucose (mg/dL) |
|---|---|
| 5.0 | 97 |
| 5.5 | 111 |
| 6.0 | 126 |
| 6.5 | 140 |
| 7.0 | 154 |
| 7.5 | 169 |
| 8.0 | 183 |
Clinical categories such as prediabetes and diabetes are defined by guideline bodies on the A1c scale (commonly 5.7 to 6.4 percent, and 6.5 percent or above), and only a clinician applies those categories to a person, using repeat testing and the wider clinical picture.
A higher than typical result means the A1c behind it was high, which usually reflects more glucose circulating over recent months, though red blood cell biology and assay quirks can also push it up.
Symptoms people report. Frequently none at all, which is the main argument for testing rather than guessing. When symptoms do appear, people describe increased thirst, waking at night to urinate, blurred vision that comes and goes, fatigue after meals, slow-healing cuts, and recurrent skin or urinary infections.
Common causes. Insulin resistance, often alongside extra visceral fat, is the most common driver in adults. Diets heavy in refined carbohydrate, low activity and low muscle mass, short or disrupted sleep, shift work and sustained stress all contribute. Separately, anything that lengthens red blood cell survival, such as iron deficiency or B12 deficiency, can raise A1c without average glucose truly being higher. Assay interference matters too. Hemoglobin variants can shift results measured on HPLC-based systems in either direction, and recognised sources of error in A1c measurement include variant hemoglobins, altered red cell turnover and chemical modifications of hemoglobin[7].
Long-term impact. Research consistently associates sustained higher glucose exposure with eye, kidney, nerve and vascular complications over years[2], while trial evidence suggests the benefit of lowering is real but has to be balanced against the risk of low-glucose episodes[3].
What is worth doing next. Treat a single high value as an observation, not a verdict. It is worth repeating the test, pairing it with fasting glucose and fasting insulin, and reviewing medications and recent illness. A high result, especially with marked thirst, weight loss or blurred vision, is worth discussing with a clinician promptly.
A lower than typical result usually means average glucose really was low, but it can also mean red blood cells are being replaced faster than normal, which makes A1c read low independently of glucose.
Symptoms people report. Many people with a low-normal result feel completely well. When genuine low glucose occurs, people describe shakiness, sweating, a racing heart, sudden hunger, irritability, light-headedness, difficulty concentrating, and sometimes night sweats or vivid dreams.
Common causes. Glucose-lowering therapy is the most frequent explanation in people already under care, and insulin therapy in particular requires close supervision, as the literature on insulin use in pregnancy illustrates, where glycaemic targets and A1c behaviour both differ from the non-pregnant state[8]. A licensed provider evaluates whether any such therapy is appropriate and adjusts it. Non-glucose explanations include haemolysis, recent blood loss or transfusion, pregnancy, and newly started iron, B12 or erythropoietin therapy, all of which shorten the average age of circulating red cells. Chronic liver disease is another, and research in people with liver cirrhosis has examined how poorly standard glycaemic markers track real glucose levels in that setting, with continuous monitoring used to fill the gap[9]. Prolonged fasting, heavy endurance exercise and alcohol can all lower glucose over shorter windows.
Long-term impact. Lower average glucose is generally associated with lower vascular risk, but low is not automatically optimal. A very low result in someone not using glucose-lowering medication more often points to red cell or liver issues worth investigating than to exceptional metabolic health. Repeated low-glucose episodes carry their own risks, a theme reinforced by trial work on tight glycaemic control in critically ill children, where tighter targets did not deliver the hoped-for advantage and raised exposure to low glucose[10].
What is worth doing next. Pair the result with a full blood count, iron studies and liver markers before assuming the average is accurate, and bring symptomatic low-glucose episodes to a clinician promptly.
Anything that changes average glucose, red blood cell lifespan, or the accuracy of the A1c assay will move the derived number.
Testing conditions. No fasting is required, and breakfast that morning will not change the value. That stability is a genuine strength. The limitation is that the result cannot tell you when during the day glucose was highest. Assay platform and laboratory calibration matter, and known sources of measurement error include hemoglobin variants and conditions that alter red cell turnover[7]. Comparing results across different labs is less reliable than tracking the same lab over time.
Medications. Glucose-lowering therapies lower it, and newer combination approaches continue to be studied, including a randomised trial of an SGLT2 inhibitor alongside a glucagon receptor antagonist in type 1 diabetes[11]. None of these are self-selected. A licensed provider evaluates whether a prescription therapy is appropriate and makes every decision about it. Iron, B12 and erythropoietin therapy can shift A1c through red cell turnover rather than through glucose.
Habits. Dietary pattern has a measurable effect. A randomised crossover trial comparing a ketogenic diet with a Mediterranean diet in people with prediabetes and type 2 diabetes found that both patterns lowered glycated hemoglobin, with differences in other markers and in sustainability[12]. Resistance training and added muscle, daily walking, sleep duration and regularity, alcohol intake and body composition changes all influence average glucose over weeks to months.
Timing. Because of red cell lifespan, meaningful change takes time to appear. Retesting a few weeks after a diet change usually shows less than you expect. Continuous glucose sensors capture variability and daily patterns that an average cannot, which is why time-in-range metrics have gained ground alongside A1c[5][6].
Estimated average glucose is reported from the hemoglobin A1c measured on FitScript's Metabolic + Insulin panel, alongside fasting glucose and fasting insulin, so the average is read next to the fasting and insulin context rather than in isolation.
Because membership includes 150+ biomarkers tested per year, the marker is re-measured on a schedule, which is what turns a single figure into a trend line. Atlas, FitScript's AI health advisor, interprets the result against the rest of the panel and writes your FitScript, describing what the pattern may reflect and what is worth raising with a clinician. Where a prescription option might be relevant, a licensed provider reviews the member and decides whether it is appropriate. No article, and no AI interpretation, substitutes for that review.
This page is educational and is not medical advice. Discuss your results and any treatment decision with a clinician or licensed provider.
The average becomes far more useful when read next to five or six related markers, each covering a different window or mechanism.
Hemoglobin A1C is the measurement the estimate is calculated from, so the two always move together. Fasting glucose shows a single morning moment and can look normal while the longer average drifts upward. Fructosamine reflects glycation over roughly two to three weeks and is sometimes used when red cell lifespan is in question. Fasting insulin often rises years before glucose does, as the body compensates for falling insulin sensitivity. HOMA-IR combines fasting glucose and insulin into a single insulin-resistance index. Uric acid is associated with insulin resistance and the wider metabolic pattern, and frequently moves in the same direction.
You can read more about each of these markers here, Hemoglobin A1C, Fasting Glucose, Fructosamine, Fasting Insulin, HOMA-IR and Uric Acid, and you can see how they are measured together on the Metabolic + Insulin panel.
Estimated average glucose is A1c expressed in mg/dL, covering roughly two to three months of glucose exposure. Typical lab reference ranges run wide, from 82 to 154 mg/dL, while FitScript's optimal range of 85 to 105 mg/dL is an optimization target rather than a diagnostic line. A higher result may reflect insulin resistance, medications, or a red cell or assay issue, and a lower result may reflect genuinely low glucose or faster red cell turnover. Trends across repeat tests, read alongside fasting glucose and fasting insulin, say far more than any single number, and the interpretation belongs with your clinician.
Calculators that use Estimated Average Glucose:
Typical lab reference ranges run from about 82 to 154 mg/dL, which corresponds to an A1c of roughly 4.5 to 7.0 percent. FitScript's optimal range is narrower at 85 to 105 mg/dL, and that is an optimization target rather than a diagnostic cut-off. Ranges vary by lab, assay method, age and sex, so read your result against the range printed on your own report and discuss it with a clinician.
Estimated average glucose is calculated directly from A1c using the formula eAG in mg/dL = 28.7 x A1c percent minus 46.7. An A1c of 5.5 percent converts to roughly 111 mg/dL, 6.0 percent to about 126 mg/dL, and 7.0 percent to about 154 mg/dL. Because it is derived arithmetic rather than a separate blood draw, it carries the same accuracy and the same limitations as the A1c behind it.
No. Fasting glucose is a single morning moment after an overnight fast, while the estimated average covers glucose exposure across the whole day and night over roughly two to three months, including post-meal rises. That is why the estimate often sits higher than a run of fasting readings. The two markers answer different questions and are most useful read together.
HPLC-based A1c methods can be affected by hemoglobin variants, which may shift the measured value in either direction, and anything that lengthens red blood cell survival, such as iron or B12 deficiency, can raise A1c without average glucose truly being elevated. An unexpectedly high result is worth repeating, ideally with the lab aware of the question, and discussing with a clinician.
Weeks to months. Because red blood cells live for roughly 120 days, the value lags behind real change, with the most recent four to six weeks weighted most heavily. Retesting too soon after a diet or training change usually understates progress. Fructosamine reflects a shorter two to three week window and is sometimes used when a faster read is needed.
It can be lower than typical, and low is not automatically better. A low value may reflect genuinely low average glucose, often related to glucose-lowering therapy, or it may reflect faster red blood cell turnover from haemolysis, recent blood loss, pregnancy, chronic liver disease or newly started iron, B12 or erythropoietin therapy. Symptoms such as shakiness, sweating or confusion are worth discussing with a clinician promptly.
No. A1c, and therefore the estimated average derived from it, does not require fasting and is not changed by what you ate that morning. Many people still have it drawn fasting because it is bundled with fasting glucose and fasting insulin on the same panel, which gives a far more complete metabolic picture than any single marker.
Not replace, complement. A sensor shows daily variability and how much time glucose spends in a target band, which an average cannot reveal. The laboratory marker remains a standardised, low-cost summary of months of exposure. Whether sensor use is appropriate is a decision for a clinician based on the individual situation.
Fasting insulin and HOMA-IR often reveal insulin resistance years before glucose markers drift, fasting glucose shows the morning moment, fructosamine covers a shorter window, and uric acid is associated with the wider metabolic pattern. A full blood count and iron studies help rule out red cell causes of a misleading value. Reviewing them as a set with a clinician is far more informative than tracking one number alone.