What is my average blood sugar if my A1c is 7?

On the ADAG equation, an A1c of 7.0% gives an estimated average glucose of 154 mg/dL (8.6 mmol/L). The important part is what comes next: the study that produced that number published a 95% interval of 123 to 185 mg/dL around it. Both ends of that range are real possibilities for one person with a 7.0% result, and the difference between them is the difference between glucose control most clinicians are happy with and control they are not.

How accurate is the A1c to average glucose conversion?

The ADAG study reported an R-squared of 0.84 and a standard deviation of prediction error of 15.7 mg/dL (0.87 mmol/L) across 507 people. Its own success criterion was that 90% of participants fall within ±15% of the predicted average, and 89.95% did — meaning roughly one person in ten fell outside even that band. The precision also gets worse as A1c rises: the paper gives error standard deviations of 13.4, 15.7 and 18.0 mg/dL at A1c values of 6, 7 and 8%.

How do I convert A1c from mmol/mol to percent?

Multiply the mmol/mol figure by 0.09148 and add 2.152. So 42 mmol/mol is 6.0%, 53 mmol/mol is 7.0% and 64 mmol/mol is 8.0%. Going the other way, subtract 2.152 and divide by 0.09148, which puts the prediabetes threshold of 5.7% at 39 mmol/mol and the diabetes threshold of 6.5% at 48 mmol/mol. This conversion comes from a head-to-head comparison of the two reference measurement systems with an r-squared of 0.998, so unlike the glucose conversion it is effectively exact.

Why does my eAG not match my glucose meter readings?

Because they are measuring different things. eAG is an average of every glucose value across roughly three months, weighted towards the most recent weeks, so it sits above a typical fasting reading and below a post-meal peak. A man whose fasting glucose is 110 mg/dL can easily have an eAG of 154 mg/dL if his readings rise after meals. NGSP also notes that fasting glucose correlates less well with A1c than afternoon and evening readings, and increasingly underestimates it as A1c rises.

Why do two A1c calculators give me different average glucose numbers?

Almost always because they are running different published equations. This page uses the 2008 ADAG equation, eAG = 28.7 × A1c − 46.7. Many older calculators and textbooks still use the 2002 DCCT equation, 35.6 × A1c − 77.3, which is steeper and returns higher numbers at every clinically relevant A1c — 243 mg/dL against 212 at an A1c of 9%, a gap of 31 mg/dL. ADAG is the more recent of the two and was built on continuous glucose monitoring rather than quarterly seven-point profiles.

When is A1c a poor guide to average glucose?

When anything shortens or lengthens the life of a red blood cell, or interferes with the assay. NIDDK names iron-deficiency anaemia, which can push A1c falsely high; sickle cell disease, blood transfusion and haemodialysis, which alter red cell lifespan; kidney failure and liver disease; and haemoglobin variants, which are more common in people of African, Mediterranean or Southeast Asian descent. The ADAG study excluded all of those conditions from its 507 participants, and also excluded children and pregnant women, so the equation on this page was never fitted to them.

What A1c counts as prediabetes or diabetes?

NIDDK gives below 5.7% as normal, 5.7 to 6.4% as prediabetes and 6.5% or above as diabetes — 39 and 48 mmol/mol respectively on the IFCC scale. In estimated average glucose those thresholds are about 117 mg/dL (6.5 mmol/L) and 140 mg/dL (7.8 mmol/L). A single reading is a screening result, not a diagnosis: confirmation normally needs a second test on a separate sample, and the bands are population cut points rather than a statement about one person.

Can I work out my A1c from my continuous glucose monitor average?

You can get an estimate, and this page's fourth chart does it — an average of 180 mg/dL inverts to an A1c of about 7.9%. But the scatter runs in this direction too, and the quantity is not identical: ADAG's average came from continuous monitoring combined with at least three days a week of seven-point fingerstick testing over three months, not from a monitor alone. If your measured A1c and your monitor average disagree persistently, that gap is itself clinically interesting and worth raising rather than explaining away.

A1c to Average Glucose Converter

Two conversions that look alike and are nothing alike

An HbA1c result measures the proportion of your haemoglobin that has sugar stuck to it. Because red blood cells live around 120 days, that proportion reflects your blood sugar over the preceding months rather than this morning. It is reported as a percentage on the NGSP scale used in the United States, and as millimoles of glycated haemoglobin per mole of total haemoglobin — mmol/mol — on the IFCC scale used across much of Europe. Both describe the same molecule.

Converting between those two scales is a calibration exercise. The two reference systems were compared head to head and the relationship published as a master equation, NGSP % = 0.09148 x IFCC + 2.152, with an r-squared of 0.998. For any practical purpose that conversion is exact: an A1c of 5.7% is 39 mmol/mol and there is no meaningful uncertainty to report.

Converting A1c into an average glucose is a completely different kind of claim. The ADAG study measured actual glucose in 507 people — roughly 2,700 readings each, over three months, from continuous monitors and fingersticks — and fitted a straight line through the result: eAG in mg/dL = 28.7 x A1c - 46.7, with an R-squared of 0.84. That line is a population average, and people scatter around it. The paper reports a standard deviation of prediction error of 15.7 mg/dL (0.87 mmol/L), and its own table gives, for an A1c of 7%, an estimate of 154 mg/dL with limits of 123 to 185.

That spread is the honest headline. An A1c of 7% is consistent with an average glucose that most clinicians would call well controlled and with one they would not, and nothing in the A1c number tells you which of the two you are. The calculator on this page therefore reports a range as well as a midpoint: the ADAG paper found that its 15% band closely matched its 90% prediction limits across the whole A1c range, and that 89.95% of individual participants fell inside it. Roughly one person in ten did not.

The estimate is also not a blood sugar reading. It is an average of every moment across about three months, weighted towards the most recent weeks, so it will sit above your fasting numbers and below your post-meal peaks. Comparing eAG with a morning fingerstick and concluding the lab is wrong is the commonest misreading of this number.

The formula

eAG (mg/dL) = 28.7 × A1c − 46.7 ADAG, Nathan 2008

eAG (mmol/L) = eAG (mg/dL) × 10 / 180.156

A1c (%) = (eAG mg/dL + 46.7) / 28.7

range = eAG ± 15% ≈ the published 90% prediction limits

NGSP (%) = 0.09148 × IFCC (mmol/mol) + 2.152

IFCC (mmol/mol) = (NGSP % − 2.152) / 0.09148

older form = 35.6 × A1c − 77.3 DCCT, Rohlfing 2002

The ADAG paper also prints an SI form, eAG (mmol/L) = 1.5944 × A1c − 2.594. It is not quite the same line as converting the mg/dL answer: it sits about 0.01 mmol/L higher, which flips the printed first decimal at two A1c values — 11.8 against 11.7 at 9%, and 13.4 against 13.3 at 10%. This page converts from mg/dL throughout so the two units can never disagree, and prints the paper’s own SI figures separately in the second chart below. NGSP’s own two web pages disagree on the same cells.

Worked example

An HbA1c of 7.4% on a US report — diabetes range, above the 6.5% threshold:

eAG = 28.7 × 7.4 − 46.7 = 165.7 mg/dL
In SI units: 165.7 × 10 / 180.156 = 9.20 mmol/L
On the IFCC scale: (7.4 − 2.152) / 0.09148 = 57.4 → 57 mmol/mol
The ±15% band: 141–191 mg/dL (7.8–10.6 mmol/L)
That band is 50 mg/dL wide. Nine men in ten with this A1c have a true average glucose somewhere inside it; one in ten is outside it altogether.
The older DCCT equation would have returned 186 mg/dL for the same A1c — 20 mg/dL higher, which is why two calculators can disagree without either being broken.

A1c to average glucose conversion chart

Find your A1c in the left column. The large figure is the estimate; the smaller one beneath is the ±15% range the ADAG study reports as its 90% prediction limits.

HbA1c (NGSP %)	eAGmg/dL	eAGmmol/L	Same A1cIFCC mmol/mol
4.0%Normal	6858–78	3.83.2–4.3	20
4.5%Normal	8270–95	4.63.9–5.3	26
5.0%Normal	9782–111	5.44.6–6.2	31
5.5%Normal	11194–128	6.25.2–7.1	37
5.7%Prediabetes range	11799–134	6.55.5–7.5	39
6.0%Prediabetes range	125107–144	7.05.9–8.0	42
6.5%Diabetes range	140119–161	7.86.6–8.9	48
7.0%Diabetes range	154131–177	8.67.3–9.8	53
7.5%Diabetes range	169143–194	9.48.0–10.8	58
8.0%Diabetes range	183155–210	10.28.6–11.7	64
8.5%Diabetes range	197168–227	10.99.3–12.6	69
9.0%Diabetes range	212180–243	11.710.0–13.5	75
9.5%Diabetes range	226192–260	12.510.7–14.4	80
10.0%Diabetes range	240204–276	13.311.3–15.3	86
11.0%Diabetes range	269229–309	14.912.7–17.2	97
12.0%Diabetes range	298253–342	16.514.0–19.0	108
13.0%Diabetes range · extrapolated	326277–375	18.115.4–20.8	119
14.0%Diabetes range · extrapolated	355302–408	19.716.8–22.7	130

Computed from eAG = 28.7 × A1c − 46.7 and NGSP % = 0.09148 × IFCC + 2.152 by the same functions the calculator above uses. Rows below 4% and above 12% are marked extrapolated: the ADAG regression was fitted between those bounds and the paper states its results "are only directly applicable to this population". The ranges are ±15%, which the paper reports as closely matching its 90% prediction limits over the full A1c range.

What the ADAG study actually published, intervals and all

Table 2 of Nathan 2008, reproduced. These eight rows are transcribed from the paper rather than computed, because the intervals come from a Bayesian model and are not symmetric about the estimate.

HbA1c (%)	eAG mg/dLwith 95% interval	eAG mmol/Lwith 95% interval	Interval widthmg/dL
5%	9776–120	5.44.2–6.7	44
6%	126100–152	7.05.5–8.5	52
7%	154123–185	8.66.8–10.3	62
8%	183147–217	10.28.1–12.1	70
9%	212170–249	11.89.4–13.9	79
10%	240193–282	13.410.7–15.7	89
11%	269217–314	14.912.0–17.5	97
12%	298240–347	16.513.3–19.3	107

Reproduced from Nathan DM, et al. Diabetes Care. 2008;31(8):1473–8, Table 2. Note the last column: the interval widens from 44 mg/dL at an A1c of 5% to 107 mg/dL at 12%, so the estimate is least precise exactly where the numbers are highest. Note also that the intervals are not centred — at 8% the paper prints 183 with limits of 147 and 217. The mmol/L column here is the paper’s own, generated from its SI equation, which is why 9% reads 11.8 and 10% reads 13.4 where converting the mg/dL figure gives 11.7 and 13.3.

IFCC mmol/mol to NGSP % conversion chart

For a UK, European or Australian lab report in mmol/mol. Unlike the glucose conversion above, this one is an assay calibration and carries no meaningful uncertainty.

IFCC (mmol/mol)	NGSP%	eAGmg/dL	eAGmmol/L
20Normal · extrapolated	3.98%	6857–78	3.83.2–4.3
25Normal	4.44%	8169–93	4.53.8–5.2
30Normal	4.90%	9480–108	5.24.4–6.0
35Normal	5.35%	10791–123	5.95.0–6.8
39Prediabetes range	5.72%	117100–135	6.55.5–7.5
42Prediabetes range	5.99%	125107–144	7.05.9–8.0
48Diabetes range	6.54%	141120–162	7.86.7–9.0
53Diabetes range	7.00%	154131–177	8.67.3–9.8
58Diabetes range	7.46%	167142–192	9.37.9–10.7
64Diabetes range	8.01%	183156–211	10.28.6–11.7
69Diabetes range	8.46%	196167–226	10.99.3–12.5
75Diabetes range	9.01%	212180–244	11.810.0–13.5
80Diabetes range	9.47%	225191–259	12.510.6–14.4
86Diabetes range	10.02%	241205–277	13.411.4–15.4
91Diabetes range	10.48%	254216–292	14.112.0–16.2
97Diabetes range	11.03%	270229–310	15.012.7–17.2
108Diabetes range · extrapolated	12.03%	299254–343	16.614.1–19.1
119Diabetes range · extrapolated	13.04%	327278–377	18.215.5–20.9
130Diabetes range · extrapolated	14.04%	356303–410	19.816.8–22.7

The NGSP master equation, NGSP % = 0.09148 × IFCC + 2.152, with r² = 0.998. The prediabetes threshold of 5.7% is 39 mmol/mol and the diabetes threshold of 6.5% is 48 mmol/mol. The NGSP column is given to two decimals deliberately: the round mmol/mol figures in NGSP's own table are themselves rounded, so 42 mmol/mol is 5.99% rather than exactly 6.00%, and the glucose columns follow the exact value rather than the tidied one. The glucose columns inherit the ADAG scatter and still carry their ±15% range; the NGSP column does not, because it does not need one.

Average glucose to A1c — the conversion in reverse

If your meter or continuous monitor reports an average, this is the A1c that average corresponds to on the ADAG line. It is not what your next lab test will necessarily say.

Average glucose	Same averagemmol/L	A1cNGSP %	A1cIFCC mmol/mol
80 mg/dL	4.4	4.4%Normal	25
90 mg/dL	5.0	4.8%Normal	29
100 mg/dL	5.6	5.1%Normal	32
110 mg/dL	6.1	5.5%Normal	36
120 mg/dL	6.7	5.8%Prediabetes range	40
130 mg/dL	7.2	6.2%Prediabetes range	44
140 mg/dL	7.8	6.5%Diabetes range	48
150 mg/dL	8.3	6.9%Diabetes range	51
160 mg/dL	8.9	7.2%Diabetes range	55
170 mg/dL	9.4	7.6%Diabetes range	59
180 mg/dL	10.0	7.9%Diabetes range	63
190 mg/dL	10.5	8.2%Diabetes range	67
200 mg/dL	11.1	8.6%Diabetes range	70
220 mg/dL	12.2	9.3%Diabetes range	78
240 mg/dL	13.3	10.0%Diabetes range	86
260 mg/dL	14.4	10.7%Diabetes range	93
280 mg/dL	15.5	11.4%Diabetes range	101
300 mg/dL	16.7	12.1%Diabetes range · extrapolated	109

The regression inverted: A1c = (eAG + 46.7) / 28.7. The same scatter applies in this direction too, and arguably matters more — a measured average glucose of 154 mg/dL predicts an A1c of 7.0%, but an individual's measured A1c can land meaningfully either side of that. A continuous monitor average is also not the same quantity ADAG fitted: its average came from monitors plus at least three days a week of seven-point fingersticks over three months.

Frequently asked questions

What is my average blood sugar if my A1c is 7?: On the ADAG equation, an A1c of 7.0% gives an estimated average glucose of 154 mg/dL (8.6 mmol/L). The important part is what comes next: the study that produced that number published a 95% interval of 123 to 185 mg/dL around it. Both ends of that range are real possibilities for one person with a 7.0% result, and the difference between them is the difference between glucose control most clinicians are happy with and control they are not.
How accurate is the A1c to average glucose conversion?: The ADAG study reported an R-squared of 0.84 and a standard deviation of prediction error of 15.7 mg/dL (0.87 mmol/L) across 507 people. Its own success criterion was that 90% of participants fall within ±15% of the predicted average, and 89.95% did — meaning roughly one person in ten fell outside even that band. The precision also gets worse as A1c rises: the paper gives error standard deviations of 13.4, 15.7 and 18.0 mg/dL at A1c values of 6, 7 and 8%.
How do I convert A1c from mmol/mol to percent?: Multiply the mmol/mol figure by 0.09148 and add 2.152. So 42 mmol/mol is 6.0%, 53 mmol/mol is 7.0% and 64 mmol/mol is 8.0%. Going the other way, subtract 2.152 and divide by 0.09148, which puts the prediabetes threshold of 5.7% at 39 mmol/mol and the diabetes threshold of 6.5% at 48 mmol/mol. This conversion comes from a head-to-head comparison of the two reference measurement systems with an r-squared of 0.998, so unlike the glucose conversion it is effectively exact.
Why does my eAG not match my glucose meter readings?: Because they are measuring different things. eAG is an average of every glucose value across roughly three months, weighted towards the most recent weeks, so it sits above a typical fasting reading and below a post-meal peak. A man whose fasting glucose is 110 mg/dL can easily have an eAG of 154 mg/dL if his readings rise after meals. NGSP also notes that fasting glucose correlates less well with A1c than afternoon and evening readings, and increasingly underestimates it as A1c rises.
Why do two A1c calculators give me different average glucose numbers?: Almost always because they are running different published equations. This page uses the 2008 ADAG equation, eAG = 28.7 × A1c − 46.7. Many older calculators and textbooks still use the 2002 DCCT equation, 35.6 × A1c − 77.3, which is steeper and returns higher numbers at every clinically relevant A1c — 243 mg/dL against 212 at an A1c of 9%, a gap of 31 mg/dL. ADAG is the more recent of the two and was built on continuous glucose monitoring rather than quarterly seven-point profiles.
When is A1c a poor guide to average glucose?: When anything shortens or lengthens the life of a red blood cell, or interferes with the assay. NIDDK names iron-deficiency anaemia, which can push A1c falsely high; sickle cell disease, blood transfusion and haemodialysis, which alter red cell lifespan; kidney failure and liver disease; and haemoglobin variants, which are more common in people of African, Mediterranean or Southeast Asian descent. The ADAG study excluded all of those conditions from its 507 participants, and also excluded children and pregnant women, so the equation on this page was never fitted to them.
What A1c counts as prediabetes or diabetes?: NIDDK gives below 5.7% as normal, 5.7 to 6.4% as prediabetes and 6.5% or above as diabetes — 39 and 48 mmol/mol respectively on the IFCC scale. In estimated average glucose those thresholds are about 117 mg/dL (6.5 mmol/L) and 140 mg/dL (7.8 mmol/L). A single reading is a screening result, not a diagnosis: confirmation normally needs a second test on a separate sample, and the bands are population cut points rather than a statement about one person.
Can I work out my A1c from my continuous glucose monitor average?: You can get an estimate, and this page's fourth chart does it — an average of 180 mg/dL inverts to an A1c of about 7.9%. But the scatter runs in this direction too, and the quantity is not identical: ADAG's average came from continuous monitoring combined with at least three days a week of seven-point fingerstick testing over three months, not from a monitor alone. If your measured A1c and your monitor average disagree persistently, that gap is itself clinically interesting and worth raising rather than explaining away.

Sources

Embed this calculator

Free to use on your own site, no permission needed and no account. The widget resizes itself to fit your layout. Keeping the credit line underneath is appreciated, but the calculator works exactly the same if you delete it.

Paste this

<!-- A1c to Average Glucose Converter — by Alpha Health Finder. Free to embed; keeping the credit line below is appreciated but not required. -->
<iframe src="https://alphahealthfinder.com/embed/tools/a1c-glucose-converter"
        title="A1c to Average Glucose Converter"
        width="100%" height="1900" loading="lazy"
        style="border:1px solid #e2e8f0;border-radius:12px;max-width:680px;width:100%"
        data-ahf-embed="a1c-glucose-converter"></iframe>
<p style="max-width:680px;margin:8px 0 0;font:13px/1.5 system-ui,-apple-system,sans-serif;color:#64748b">
  <a href="https://alphahealthfinder.com/tools/a1c-glucose-converter">A1c to Average Glucose Converter</a> by <a href="https://alphahealthfinder.com">Alpha Health Finder</a>
</p>
<script src="https://alphahealthfinder.com/embed/resize.js" async></script>

Next steps

Get an HbA1c without a clinic visitProviders that run HbA1c alongside fasting glucose and insulin from one draw — which is what you need to tell a high A1c from a misleading one.What to test alongside A1cFasting insulin, a lipid panel and a full blood count between them explain most of the cases where an A1c does not match how someone feels.