What is MARD for a CGM?

MARD stands for Mean Absolute Relative Difference. It is a summary statistic that describes, on average, how far a CGM's readings sit from a reference glucose measurement, expressed as a percentage. A lower MARD generally indicates readings that track the reference more closely, but MARD is a single average and does not capture every situation.

Does a CGM replace a lab or blood test?

No. A CGM estimates glucose in interstitial fluid and is not a perfect substitute for a laboratory blood test. It is excellent at showing trends and patterns over time, but for diagnosis and certain treatment decisions, clinicians rely on validated lab or point-of-care blood measurements.

When should I confirm a CGM reading with a finger-stick?

Confirm with a finger-stick when a reading does not match how you feel, when glucose is changing rapidly, during a suspected low, or when the device prompts you to. Always follow your device's labeling and your clinician's guidance about when finger-stick confirmation is required.

CGM basics · accuracy

How Accurate Is a CGM? MARD, Explained

6 min read · Updated July 2026

Continuous glucose monitors have become remarkably good, but "accurate" is a more nuanced idea than a single number suggests. Accuracy depends on the device, on how high or low your glucose is, and even on which day of a sensor's life you check. This article explains the concept most often used to describe CGM accuracy — MARD — and when it still makes sense to reach for a finger-stick. For the bigger picture, see the CGM guide.

What MARD means

MARD stands for Mean Absolute Relative Difference. In plain terms, it answers a simple question: on average, how far off are a sensor's readings from a trusted reference measurement taken at the same moment? For each paired reading, you calculate the difference as a percentage of the reference value, ignore whether it is high or low (that's the "absolute" part), and then average across many readings.

The result is a single percentage. A lower MARD generally means the readings sit closer to the reference, and a higher MARD means they scatter further away. Because it rolls thousands of comparisons into one figure, MARD is a convenient shorthand — but that convenience is also its limit. An average cannot tell you how a device behaves in the specific moment that matters to you.

MARD summarizes how far readings fall from a reference, on average. It is a useful headline figure, but it flattens the moment-to-moment variation shown by the individual gaps.

Why accuracy varies

No sensor performs identically in every condition, and a headline MARD figure can hide meaningful variation. Several factors commonly influence how close a reading is:

The device. Different CGM systems use different sensor chemistry, algorithms, and calibration approaches, so their accuracy profiles differ. We compare the categories in CGM sensors compared.
The glucose level. A sensor may track differently in the low, normal, and high ranges. Performance during hypoglycemia is especially important and is often reported separately.
The sensor day. Accuracy can shift over a sensor's wear period — early on during warm-up, and sometimes near the end of its life. How a CGM works, including warm-up, is covered in how a CGM works.
Rate of change and lag. Because a CGM reads interstitial fluid, it can trail blood glucose when levels move quickly, which shows up as apparent error during rapid rises or falls.

Because of all this, the most reliable accuracy information for any specific product comes from its manufacturer labeling and the regulatory clearance documents, not from a single number quoted out of context.

A CGM is not a perfect lab substitute

It is worth stating plainly: a CGM estimates glucose in interstitial fluid and is not a flawless stand-in for a laboratory blood test. That is not a criticism — it is what the technology is designed to do. Where CGM shines is showing the direction, speed, and pattern of glucose across the whole day, information a single lab draw cannot provide. Where a validated blood test still leads is in the precise, single-point measurements used for diagnosis and certain clinical decisions.

The two are complementary rather than interchangeable. This is also why average-based summaries derived from CGM, such as the Glucose Management Indicator, are described as estimates that approximate — but do not exactly equal — a laboratory HbA1c.

When to confirm with a finger-stick

Modern CGMs let many people make day-to-day decisions from sensor data, but there are moments when a finger-stick blood test is the sensible check:

When a reading doesn't match how you feel — for example, the sensor looks fine but you have symptoms of a low.
When glucose is changing rapidly and the lag could mean the number hasn't caught up.
When you are treating a suspected low and need to act on a trustworthy value.
Whenever your device's instructions or your clinician tell you a confirmation is required.

None of this undermines the value of CGM; it simply reflects good practice. For a fuller comparison of the two methods and their strengths, see CGM vs finger-stick testing. Clinical software such as Endobits works from CGM data to support a clinician's interpretation — as decision support under professional oversight, never as a diagnosis by itself.

Understand your glucose in context

Trends over time often say more than any single reading. See how continuous data can add insight to your metabolic picture.

Check your glucose

Sources

U.S. Food and Drug Administration, Glucose Testing Devices. National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), Continuous Glucose Monitoring. American Diabetes Association, Devices & Technology.

This article is educational and not medical advice. Accuracy claims for any specific device are defined in its manufacturer labeling and regulatory clearance; consult those and a qualified healthcare provider. Endobits is clinical decision-support software used under clinician oversight, not a diagnostic device.