Is Dexcom Stelo Accurate? What Its FDA Filing Shows
The first over-the-counter continuous glucose monitor cleared by the FDA, no prescription required. Here's what's actually verified about it.
Stelo is Dexcom's glucose sensor built for people who don't have diabetes and aren't on insulin, sold directly to consumers rather than through a doctor. It's the same underlying sensor category as Dexcom's prescription G7, adapted for an over-the-counter market. Below is what's actually verified about what it measures, how it's cleared, and where the evidence on non-diabetic CGM use stands, separate from marketing claims made by comparison sites that don't cite a primary source.
What Does Stelo Actually Measure?
Stelo is a transcutaneous sensor worn on the back of the upper arm that measures interstitial glucose (glucose in the fluid between cells, a close proxy for blood glucose) continuously, displaying a glucose graph, trend arrows, and Time in Range, the clinical-standard metric for what percentage of the day glucose stays within a target band.
| Sensor wear duration | 15 days per sensor |
| Intended population | 18 years and older, not on insulin |
| Prescription required | No (over-the-counter) |
| Price | $99 per sensor (one-time purchase), HSA/FSA eligible |
Dexcom's own materials note that roughly 78 percent of sensors last the full 15-day wear period, meaning about one in five doesn't. Stelo is explicitly not recommended for anyone with problematic hypoglycemia, and Dexcom's own guidance is not to take medical action based on a Stelo reading alone without consulting a clinician.
Is Stelo FDA Cleared?
Yes. The FDA cleared the "Stelo Glucose Biosensor System" (applicant: Dexcom, Inc.) on March 5, 2024, as a traditional 510(k) clearance, the first over-the-counter continuous glucose monitor cleared for sale without a prescription in the U.S. The FDA later cleared it for use by children aged 2 and up who aren't on insulin.
FDA clearance establishes that the device met the agency's safety and performance requirements for its stated intended use. It does not mean the device has been independently validated for the specific question of interpreting glucose patterns in a healthy, non-diabetic adult using it out of general curiosity, which is a separate, newer research question covered below.
How Accurate Is Stelo?
Dexcom publishes that 93 percent of Stelo readings above 70 mg/dL fall within 20 percent of a laboratory reference. What its FDA filing adds is who that was measured in: Dexcom ran a clinical study of Stelo itself, and the participants were adults with diabetes. Stelo is sold to people without diabetes, and no accuracy result from that population appears in the filing.
Correction, 2026-09-19
This section previously said no specific accuracy percentage was traceable to Dexcom's own primary source. That undersold what's actually published: Dexcom's own site states a 93 percent within-20/20%-YSI figure. It's a different metric than MARD, so any MARD percentage circulating for Stelo on comparison sites still isn't confirmed against a Dexcom primary source and shouldn't be treated as equivalent to this one.
What Dexcom's FDA filing actually says
Stelo was cleared under 510(k) number K234070 with the Dexcom G7 as its predicate device. The filing states Stelo "shares identical hardware, material, chemical composition, principle of operation and energy source" with the G7, so it is the G7 sensor sold over the counter.
Unlike Abbott's Lingo, which relied on the clinical data of an earlier prescription sensor, Dexcom ran a clinical study on Stelo itself, comparing its readings against venous plasma measured on a YSI laboratory analyser across the full measuring range over a 15-day wear period. That study was conducted in adults aged 18 and over with diabetes, and the filing states the device met the FDA's special controls for this device class. Reported device-related adverse events were local infection, skin irritation and pain or discomfort.
A practical difference the filing documents: a 15-minute delay
The Stelo transmitter sends a value to the app every 5 minutes, but the app presents readings to the user every 15 minutes. The filing states plainly that the most recent displayed glucose value might be up to 15 minutes old. Abbott's Lingo updates every minute. For watching a fast post-meal rise, that difference is material.
What published research says about CGMs in people without diabetes
There is no independent published test of Stelo in people without diabetes. The limited independent evidence on CGMs in healthy people points the same way for both brands. Under simulated altitude, an earlier-generation Dexcom sensor, the G6 rather than the G7 that Stelo uses, had a mean absolute relative difference of 41.8 percent in healthy volunteers against 17.5 percent in people with diabetes, so it was markedly less accurate in the healthy group under that condition (Diabetes Technology & Therapeutics, 2026). In a separate crossover trial of 15 healthy adults, a continuous glucose monitor read about 0.9 mmol/L higher than fingerstick samples and overstated time above 7.8 mmol/L roughly fourfold (American Journal of Clinical Nutrition, 2025). Neither study tested Stelo, and the first tested a different Dexcom generation.
Can acetaminophen (Tylenol) throw off a Stelo reading?
Stelo's own filing does not list interfering substances. The filing for the G7, the sensor Stelo shares its chemistry with, lists two: hydroxyurea, and acetaminophen at more than 1 gram every 6 hours in adults (FDA 510(k) K213919).
The reason the warning exists is visible in older sensors. In 10 healthy adults given 1 gram of acetaminophen, Dexcom Seven Plus, Dexcom G4 Platinum and Medtronic Guardian sensors read anywhere from about 85 to 400 mg/dL while blood glucose held near 90, and the false rise followed acetaminophen levels in the fluid under the skin (Diabetes Technology & Therapeutics, 2016). A follow-up in 19 healthy adults wearing Guardian and G4 Platinum sensors found some evidence of interference from lisinopril, albuterol, acetaminophen, atenolol and red wine, and its authors called for the same tests on newer sensors (Journal of Diabetes Science and Technology, 2017). Both tested generations before the G7, so they show why the warning exists rather than how large the effect is on Stelo.
What Is a Normal Stelo Reading?
For someone without diabetes, most of a CGM day falls between 70 and 140 mg/dL. Healthy, non-obese volunteers wearing a Dexcom G6, the generation before the G7 sensor inside Stelo, spent a median of 96 percent of the day in that range, averaged 98 to 99 mg/dL (104 over age 60), and spent about 30 minutes a day above 140 (Journal of Clinical Endocrinology and Metabolism, 2019). Framingham Heart Study adults with normal blood sugar, average age 58.5, spent 87 percent of the day in range and about 3 hours above 140 (Journal of Clinical Endocrinology and Metabolism, 2025 [funding listed from Dexcom]). KnowYourPrime's CGM desk rates this normal range as Established.
What a normal meal looks like
In the same healthy-volunteer group, glucose rose from about 93 mg/dL before a meal to an average peak near 130, about an hour and a half later, and exercise lowered it by about 15 mg/dL (Journal of Diabetes Science and Technology, 2021 [conflict statement: an author has Dexcom research grants and sits on its advisory board]). Stelo shows a new value every 15 minutes, and Dexcom's FDA filing says the displayed value may be up to 15 minutes old, so the peak on screen can lag the real one.
Where normal ends is not settled
Framingham adults with prediabetes spent 77.1 percent of the day in the 70 to 140 range against 87 percent with normal blood sugar (Journal of Clinical Endocrinology and Metabolism, 2025 [funding listed from Dexcom]), but there is no agreed CGM threshold for a healthy person. When 18 CGM specialists reviewed reports from people without diabetes, they agreed poorly on who needed follow-up (Journal of Diabetes Science and Technology, 2026 [conflict statement: authors have received fees or research support from Dexcom and Abbott]).
What Does the Research Actually Say About Non-Diabetic CGM Use?
Separately from Stelo as a specific product, a growing body of research looks at what continuous glucose data means for people without diabetes: whether the glucose spikes a CGM shows in a healthy person carry the same significance they would in someone with impaired glucose control, and whether wearing a CGM changes eating or exercise behavior just by being worn.
That evidence base is real but still developing, newer and thinner than the decades of diabetes-management CGM research it's often compared against.
What Stelo Doesn't Do
Real, structural limitations tied to what a single-purpose glucose sensor is built for.
No other wearable metrics
Stelo is a glucose sensor only. It doesn't track HRV, sleep, steps, or heart rate the way a watch or ring does; Oura's app is the one live integration on this site pairing Stelo's data with a wrist/ring device's own metrics.
Not for insulin users or diagnosed diabetics
Stelo's clearance and Dexcom's own guidance are specific to people 18 and up (or 2 and up for the pediatric clearance) who are not on insulin. It is a wellness product, not a diabetes-management device.
Common questions
Is Dexcom Stelo accurate?
Dexcom reports that 93 percent of Stelo readings above 70 mg/dL fall within 20 percent of a laboratory reference. Its FDA clearance rests on a clinical study of Stelo in adults with diabetes, measured against venous plasma on a YSI analyser. No accuracy study in people without diabetes, the people Stelo is sold to, appears in the FDA filing, and no independent study has tested Stelo in that group.
What sensor does Stelo use?
Stelo uses the Dexcom G7 sensor. Its FDA filing names the G7 as the predicate device and states Stelo shares identical hardware, materials, chemistry and principle of operation with it. Each sensor lasts up to 15 days with a 12-hour grace period.
How often does Stelo update?
The Stelo app shows a new glucose reading every 15 minutes, even though the sensor sends a value every 5 minutes. Dexcom’s FDA filing states the most recent displayed value might be up to 15 minutes old. Abbott’s Lingo updates every minute.
Is Stelo or Lingo more accurate?
No study has compared them directly. Lingo now has one independent test in people without diabetes: in 39 adults without diabetes, Lingo differed from fingerstick blood by 12.9 percent on average, with 66.5 percent of readings within 15 percent, and agreement was worse after a sugary drink than after a placebo, 15.3 against 10.6 percent (Prakash et al., Journal of Diabetes Science and Technology, 2026). Stelo has none in that group. Stelo was cleared on a clinical study of Stelo itself in adults with diabetes. Lingo was cleared on clinical data from the FreeStyle Libre 2, also collected in people with diabetes, with no separate Lingo study. Stelo is built on the Dexcom G7 and Lingo on the FreeStyle Libre 2.
Correction, 2026-10-15
The answer on Stelo versus Lingo previously said neither had published accuracy data in people without diabetes. A 2026 independent study tested Lingo in that group; the answer has been corrected.
What is the normal range on Stelo?
Stelo itself does not define a normal range for people without diabetes. Reference studies on Dexcom sensors found healthy volunteers spent a median of 96 percent of the day between 70 and 140 mg/dL, and middle-aged adults with normal blood sugar spent 87 percent, with average glucose around 98 to 99 mg/dL in healthy volunteers.
Is 140 blood sugar high after eating on a CGM?
Not for a short time. Middle-aged adults with normal blood sugar spent about 3 hours a day above 140 mg/dL on a CGM, and healthy volunteers about 30 minutes. In healthy volunteers the average meal peak was close to 130 mg/dL.
Stelo's real differentiator is showing Time in Range directly, unmodified, the same framework a clinical CGM user would already recognize, rather than abstracting it into a proprietary score the way Abbott's Lingo does. What it can't offer yet is a MARD figure comparable to other CGMs' published numbers, or a settled research base on what its data means for someone who isn't managing a diagnosed condition.