This content is for informational purposes only. Always consult your healthcare provider before making health decisions. CGM accuracy data comes from manufacturer-sponsored clinical studies — real-world accuracy may vary.
MARD figures are sourced from peer-reviewed pivotal trials where available (Libre 3, G7) and from FDA 510(k) clearance submissions for OTC biosensors (Lingo, Stelo), which report MARD as a range rather than a single figure, as of July 2026. Study conditions, reference methods, and populations vary across devices — direct comparison should be made with this in mind.
What MARD means — and why it matters
MARD stands for Mean Absolute Relative Difference. It is the standard accuracy metric for CGM devices, calculated by comparing CGM readings to a reference blood glucose measurement (typically a laboratory glucose analyzer) across many paired readings during a clinical trial.
The formula is simple: for each paired reading, the absolute percentage difference between the CGM reading and the reference is calculated. MARD is the average of those percentage differences across the full study dataset.
Lower MARD = more accurate. A MARD of 7.8% means the device's readings differ from the reference by 7.8% on average. A device with 7.8% MARD is more accurate than one with 9.0% MARD.
MARD is the most widely used accuracy metric because it is scale-independent — it doesn't matter whether glucose is measured in mg/dL or mmol/L. It is also the metric required for FDA clearance submissions.
Important limitation: MARD is reported from controlled clinical studies. Real-world accuracy depends on additional factors including sensor placement, physical activity, altitude, and individual variation.
Compare accuracy and costs across all major CGMs.
Published MARD by device — 2026 data
FreeStyle Libre 3 Plus (Abbott's newer 15-day variant) is not shown — its MARD has not been independently published as distinct from the original Libre 3's figure above.
FreeStyle Libre 3 — 7.8% MARD
The FreeStyle Libre 3 posts a 7.8% MARD in Abbott's published pivotal trial data, making it one of the most accurate prescription CGMs available. It uses the same core sensing platform as the Libre 2 but with a significantly slimmer profile and real-time Bluetooth streaming. Requires a prescription. Abbott's newer FreeStyle Libre 3 Plus extends wear to about 15 days; its MARD has not been independently confirmed as distinct from this 7.8% figure. Read full review →
Abbott Lingo — 8.1–10.7% MARD
The Abbott Lingo, Abbott's OTC device, posts an 8.1–10.7% MARD — a higher (less accurate) range than Dexcom Stelo's published 7.3–9.1%, though the ranges overlap and are not directly comparable across separate trials. The Lingo is built on the same FreeStyle Libre sensing platform as the Libre 3 and is the cheapest OTC option at $89/month. No prescription required.
Dexcom G7 — 8.2% MARD (10 Day) / 8.0% MARD (15 Day)
The original 10-day Dexcom G7 posts an 8.2% overall MARD, with notably strong performance in the hypoglycemic range (below 70 mg/dL) — a critical accuracy zone for insulin users. Dexcom's newer G7 15 Day variant reports 8.0% MARD and roughly 15.5-day wear including its grace period, and is now the recommended variant for most adults. The G7 is a prescription device and the clinical standard for users managing insulin-treated diabetes. Read full review →
Dexcom Stelo — 7.3–9.1% MARD
The Dexcom Stelo, Dexcom's OTC offering, posts a 7.3–9.1% MARD — a lower (better) range than Abbott Lingo's published 8.1–10.7%, though the ranges overlap and figures from different manufacturers' trials are not strictly comparable. It shares the G7's sensor technology but is configured for non-insulin-using wellness users. Accuracy is strong for the OTC category and wellness use cases.
Full accuracy comparison table
| Device | MARD | Sensor life | Monthly cost | Rx required |
|---|---|---|---|---|
| Dexcom G7 15 Day | 8.0% | ~15.5 days | Check price | Yes |
| Abbott Lingo | 8.1–10.7% | 14 days | Check price | No |
| FreeStyle Libre 3 | 7.8% | 14 days | Check price | Yes |
| FreeStyle Libre 3 Plus | Not independently published | ~15 days | Check price | Yes |
| Dexcom G7 (10 Day) | 8.2% | 10 days | Check price | Yes |
| Dexcom Stelo | 7.3–9.1% | 15 days | Check price | No |
MARD figures from manufacturer pivotal trial publications. Monthly costs are retail estimates as of May 2026. Prescription device costs may vary significantly with insurance coverage.
Accuracy isn't the whole picture — it doesn't tell you what you'll pay, whether a device needs a prescription, or which plan fits how you'll use it. For OTC access, pricing, and plan details, see our Stelo cost and plans breakdown and Stelo vs Abbott Lingo comparison.
See what the most accurate CGMs cost for 30, 90, or 365 days.
What affects real-world CGM accuracy
Sensor placement
Approved wear sites vary by device and age group — check the specific device's instructions before placement. Dexcom G7 and G7 15 Day are approved for the back of the upper arm (ages 7+), or upper buttocks as an alternative for ages 2–6; unlike the earlier G6, the abdomen is not an approved G7 site in the US. Consistent placement in the manufacturer-recommended zone produces the most reliable readings, and placing the sensor on fatty tissue reduces variability.
Compression lows
Sleeping on a sensor can compress the tissue, reducing interstitial fluid flow and producing falsely low readings. This is a hardware limitation of all current CGM designs, not specific to any device.
Calibration
Both Dexcom and Abbott OTC devices are factory-calibrated — no fingerstick is required. Factory calibration eliminates user calibration error, a significant accuracy advantage over older CGM generations.
Altitude
High altitude can affect sensor accuracy for some CGM platforms. Manufacturer guidance typically notes altitude limitations — usually above 10,000 feet — in device documentation.
Extreme temperatures
Very high or very low ambient temperatures can temporarily affect sensor accuracy. All devices should be stored and worn within manufacturer-specified temperature ranges.
Rapid glucose changes
Interstitial glucose lags behind blood glucose by 5–15 minutes. During periods of rapid change — right after meals or during intense exercise — the CGM reading may differ meaningfully from a simultaneous blood glucose measurement.
Accuracy during exercise versus rest
CGM accuracy during exercise has been studied in several device-specific publications. The general finding is that all devices are somewhat less accurate during high-intensity exercise, due to the lag between blood glucose and interstitial glucose being amplified during rapid glucose change.
The Dexcom G7 has published accuracy data specifically during exercise, showing maintained MARD performance within acceptable ranges during moderate exercise. The Abbott Lingo and FreeStyle Libre 3, based on the same platform, show similar performance characteristics.
For athletes who want CGM data during training, all devices listed provide useful directional trend data. The absolute accuracy of any point-in-time reading during intense exercise should be interpreted with the interstitial lag in mind.
Which device is most accurate?
Abbott Lingo — cheapest OTC CGM
At 8.1–10.7% MARD, the Abbott Lingo's published range is higher (less accurate) than Dexcom Stelo's 7.3–9.1% — though the ranges overlap and figures from separate manufacturer trials aren't directly comparable. Lingo's case for non-prescription users rests on price ($89/month, the lowest OTC cost) and the FreeStyle Libre sensor platform, not on leading accuracy.
Read full Lingo review →Dexcom G7 15 Day and FreeStyle Libre 3 — closest published figures
Dexcom's newer G7 15 Day (8.0% MARD) and the FreeStyle Libre 3 (7.8% MARD) currently report the closest, best-documented single MARD figures among prescription CGMs — both ahead of the original 10-day G7 (8.2%). FreeStyle Libre 3 Plus does not yet have an independently confirmed MARD figure of its own. For users who need a prescription CGM, Libre 3 combines strong accuracy with 14-day wear at a lower retail price than either G7 variant. The G7 has stronger accuracy in the hypoglycemic range specifically.
Read full Libre 3 review →Compare all CGM costs for any monitoring period.