MARD — Mean Absolute Relative Difference — is the standard accuracy metric for CGM devices. A lower MARD means a smaller average difference from the reference in that study. This page lists published pivotal trial figures for the CGMs available in the US in 2026; because protocols differ, the order is for reference and is not a clinical ranking.
Medical Disclaimer: This content is for informational purposes only. Always consult your healthcare provider before making health decisions.
This comparison quotes study figures for specifically named sensor models. Do not transfer a figure from the original Libre 3 to a Plus variant or to another sensor; consult current manufacturer documentation for the model you will use.
A range and a pooled average are different summaries. Read how each study defines and reports its results; neither summary establishes a clinical ranking across separate protocols.
| Device | MARD | Rx required | Monthly cost | Source |
|---|---|---|---|---|
| Dexcom G7 15 Day | 8.0% | Yes | Check price | Dexcom pivotal trial (2025) |
| FreeStyle Libre 3 | 7.8% | Yes | Check price | Abbott pivotal trial (Alva et al., Diabetes Ther 2023; ages 6+) |
| FreeStyle Libre 3 Plus | 8.2% adults / ages 6–17 (manufacturer, YSI); 9.7% ages 2–5 (BGM) | Yes | Check price | Abbott US formulary kit, September 2025, page 5 |
| Dexcom G7 (10 Day) | 8.2% adult arm | Yes | Check price | Dexcom pivotal trial |
| Medtronic Guardian 4 | 10.6% adult arm / 11.6% ages 7–17 arm | Yes | Varies | FDA P160017/S091, table 17 |
| Eversense 365 | 8.8% | Yes + procedure | Varies | ENHANCE study (Bailey et al. 2025) |
| Eversense E3 | 8.7% (US IFU, SW605) | Yes + procedure | Varies | US E3 IFU SW605; 12,154 pairs |
| Dexcom Stelo | 7.3–9.1% (historical adults, by wear period) | No (OTC) | Check price | Dexcom pivotal trial |
| Abbott Lingo | 8.1–10.7% (range, by glucose range) | No (OTC) | Check price | FDA K233655, six selected Libre 2 bins, 55–200 mg/dL |
MARD is the mean absolute relative difference between sensor readings and reference measurements in a particular study. It is an average across paired readings, not an error bound for each reading or a direct clinical ranking across separate studies.
A lower MARD means a smaller average difference in that study. MARD depends on the study protocol, the reference method, the glucose range and the number of paired readings, so a 7.9% figure from one study and an 8.7% figure from another do not show that one device reads closer to the true value. A small difference between separate studies is not by itself a clinical advantage.
Different trial populations and reporting methods: Abbott Lingo's FDA submission reports an 8.1–10.7% MARD range (by glucose range), based on the same pivotal sensor data as the Libre 2 platform, studied in people with diabetes; FreeStyle Libre 3's 7.8% single-figure MARD comes from its own pivotal trial, also in people with diabetes. A range and a single average are not directly comparable, and neither is directly comparable to Dexcom's figures, which come from separate trials with their own populations and methods.
Accuracy can vary with the glucose range and the study conditions. An overall MARD figure does not establish performance for an individual low-glucose reading or clinical superiority over another device.
Real-world vs clinical: MARD from pivotal trials may differ from real-world performance due to sensor placement, activity, temperature, and individual variation.
MARD, sensor life, cost, and clinical notes — all in one place.
Published MARD figures come from separate manufacturer studies and do not establish which device is more accurate. Choose a device based on its intended use and the features you need with advice from your healthcare professional, and check manufacturers for current prices.
FreeStyle Libre 3 (7.8% MARD) — available with a prescription; price depends on your plan or pharmacy. Reads every minute, lasts 14 days.
Published MARD figures come from separate manufacturer studies and do not establish which device is more accurate. Choose a device based on its intended use and the features you need with advice from your healthcare professional, and check manufacturers for current prices.
Enter your own prices to compare calculated cost for your period.
Accuracy, cost, sensor life, and what each device is cleared for — in one comparison.
Compare all CGM devices →Medical source review, 25 October 2026. Sources apply to their named models, populations and endpoints: Stelo FDA K260935; Original Libre 3, Alva 2023; Libre 3 Plus manufacturer kit, page 5; Guardian 4 FDA, table 17; Eversense E3 US IFU; Eversense 365 ENHANCE; Eversense 365 US guide; Eversense 365 and twiist.