CGM Accuracy
CGM Accuracy Explained: What MARD Means & How Accurate Sensors Are (2026)
By GlucoRx Editorial Team · Last updated 14 September 2026
MARD — Mean Absolute Relative Difference — is the main number used to describe how accurate a continuous glucose monitor is. It is the average percentage difference between the CGM's readings and paired reference (blood) glucose values: the lower the percentage, the more closely the CGM tracks the reference.
A MARD below about 10% is generally regarded as good, and most modern CGMs sit around 8–10%. There is no official regulatory MARD “pass mark” — under-10% is a widely-used rule of thumb, not a formal standard. And because MARD figures come from different studies with different methods, a gap of a few tenths of a percent between two sensors is not meaningful. This guide is general information, not medical advice.
What is MARD, and how is it measured?
MARD stands for Mean Absolute Relative Difference. It is calculated by taking many paired readings — the CGM value and a reference glucose value measured at the same moment — working out the absolute percentage difference for each pair, and averaging them. A MARD of 9%, for example, means the CGM's readings differ from the reference by an average of 9%.
The reference is usually a laboratory analyser (often a YSI instrument) or a fingerstick/venous blood test, taken under a defined study protocol. Because the reference and the conditions differ between studies, the same sensor can post different MARD numbers in different studies.
What counts as a good MARD?
As a rule of thumb, a MARD below roughly 10% is considered good and clinically useful, and today's leading CGMs report figures around 8–10%. Importantly, there is no single official regulatory MARD cut-off for consumer CGMs — “under 10%” is a widely-cited guideline, not an ISO or FDA pass/fail line.
MARD is also only part of the picture. How a sensor performs in the low range, how much it lags during rapid changes, and how consistent it is day to day all matter alongside the headline number.
Why you can't rank CGMs by MARD
MARD figures from different studies are not directly comparable. The reference method, the number of paired points, the glucose ranges covered, how fast glucose was changing, and the patient population all shift the result. That is why the same device can report different MARDs in different studies.
In practice this means a difference of a few tenths of a percentage point between two sensors — say 8.2% versus 8.66% — is not a meaningful ranking. Read the numbers below as a band that today's CGMs fall into, not a league table.
MARD of CGMs sold in India (read as a band, not a ranking)
| Device | MARD | Basis |
|---|---|---|
| Dexcom G7 | 8.2% | Peer-reviewed (Garg et al., 2022; adults, arm) |
| FreeStyle Libre 2 Plus | 8.2% | Manufacturer / ADA Consumer Guide |
| GlucoRx Vixxa 2 | 8.66% | Manufacturer clinical data |
| Sinocare iCan i3 | 8.71% | Manufacturer |
| FreeStyle Libre 2 (original 14-day) | 9.2% | Peer-reviewed |
All of these sit within about half a percentage point, and they are drawn from different studies with different protocols and populations — so the figures are not directly comparable and should be read as a band, not a ranking. Manufacturer-reported figures have not necessarily been independently peer-reviewed.
MARD vs finger-prick meters (and the ISO 15197 standard)
You may see the standard “ISO 15197:2013” mentioned around glucose accuracy. It is the accuracy standard for fingerstick blood glucose meters, not for CGMs. It requires at least 95% of a meter's results to fall within ±15 mg/dL of the lab reference below 100 mg/dL (and within ±15% at or above 100 mg/dL), plus a separate criterion that at least 99% of results fall within zones A and B of the consensus error grid.
Fingerstick meters typically achieve a MARD of about 3–6%, lower than a CGM's 8–10%. That does not mean a CGM “fails” a standard — ISO 15197 does not apply to CGMs, which measure interstitial fluid rather than blood and are assessed differently. A CGM's value is continuous data and alerts, not matching a lab meter reading-for-reading. See our CGM vs finger prick guide for more.
How to sanity-check your own sensor
If you want to check a sensor against a finger-prick, do it when your glucose is stable — not right after a meal, exercise or a low, when the natural interstitial lag makes the two legitimately differ. A single close match or small gap is expected; readings won't be identical.
If your CGM is consistently far from your meter across several stable checks, contact the manufacturer's support — a sensor can occasionally be faulty. A CGM is a monitoring aid; follow your doctor's advice for treatment decisions.
Frequently Asked Questions
What does MARD mean?
MARD (Mean Absolute Relative Difference) is the average percentage difference between a CGM's readings and paired reference blood glucose values. A lower MARD means the CGM tracks the reference more closely; a MARD of 9% means readings differ from the reference by about 9% on average.
What is a good MARD for a CGM?
A MARD below roughly 10% is generally considered good, and modern CGMs report around 8–10%. There is no official regulatory MARD cut-off for consumer CGMs — “under 10%” is a rule of thumb, not a formal pass/fail standard.
What is the MARD of the GlucoRx Vixxa 2?
8.66%, as per the manufacturer's clinical data. We could not find an independent peer-reviewed MARD study for the Vixxa 2, so treat it as a manufacturer-reported figure. It sits within the same ~8–9% band as other CGMs sold in India, and cross-study MARD figures are not directly comparable.
Do CGMs have to meet ISO 15197?
No. ISO 15197:2013 is the accuracy standard for fingerstick blood glucose meters, not CGMs. CGMs measure interstitial fluid and are assessed differently, which is one reason their MARD (~8–10%) is higher than a meter's (~3–6%).
Important information & sources
AI-assisted content. This page was created with the help of AI and public sources, and reviewed against them, as of 14 September 2026. Details may change — please verify current specifications and prices with the manufacturer or an authorised seller before you rely on them.
This page is published by GlucoRx, the manufacturer of the Vixxa 2 sensor. It is for general information only and is not medical advice. Individual results vary. Consult your doctor or diabetes care specialist before making changes to your monitoring or treatment.
MARD definitions, standards and device figures were retrieved in September 2026 from the sources below. Device MARDs are drawn from separate manufacturer and peer-reviewed studies and are not directly comparable.
All trademarks are the property of their respective owners. Dexcom is a trademark of Dexcom, Inc.; FreeStyle Libre is a trademark of Abbott; Sinocare and iCan are trademarks of Sinocare. If you spot an error on this page, please email info@glucorx.co.in and we will correct it promptly.
Try Vixxa 2 — ₹3,099–₹3,399/sensor
CDSCO-licensed. 15-day wear. IP68 waterproof. Free shipping across India.