A flat glucose line doesn’t prove your fueling is fine. A dip under 70 mg/dL doesn’t prove you’re about to bonk. A continuous glucose monitor (CGM) reads sugar in the fluid under your skin, not the fuel stored in your muscles. That gap explains most of the confusion.
The sensor is a small patch you wear on your arm or belly. It reports a glucose number all day. Type 1 diabetics rely on it. Athletes without diabetes now buy it to tune their carbs per hour.
Does it help? Sometimes. Here’s what the data says.
How Accurate Is a CGM During Exercise?
Sensors are good at rest and worse when you move. Accuracy is scored by MARD (mean absolute relative difference), the average gap between the sensor and a finger-prick reference. Lower is better.
In a 13-person para-cyclist study, the Abbott Libre Sense sensor had a MARD of 12% at rest. During exercise it jumped to 34%. Most of the error was overestimation. In short: the same sensor went from “pretty close” to “off by a third” once the riders started pedaling.
Studies in adults with type 1 diabetes point the same way. A 2021 review of sensor accuracy during exercise found errors climb when glucose moves quickly. We don’t have a large dataset for healthy athletes yet.
Then there’s lag. Sensors read the fluid between your cells, not your blood. Reviews put the delay near 5 minutes at rest and up to 12 to 24 minutes when glucose changes fast. Think of it like a rearview mirror. You see where you were, not where you are.
Your glucose can crash and recover before the sensor tells you.
What a CGM Actually Measures
Glucose is a tiny pool. A 70 kg person has about 4 g circulating in the blood. Muscle and liver hold far more, and that’s what carries you through a long session.
Picture a car. Blood glucose is the small cup of fuel sitting in the carburetor. Glycogen is the tank. The body defends that cup hard, so it can look steady while the tank runs low. Muscle glycogen predicts fatigue better than blood glucose does.
A CGM watches the cup. Your race depends on the tank.
Two Elite Athletes, Two Opposite Curves
The 2026 case series by Skroce and colleagues followed three elite athletes without diabetes during record events. Two of them tell the story. (Several authors disclosed ties to the CGM industry.)
| Relay cyclist | Everesting cyclist | |
|---|---|---|
| Event | 44 h 20 min relay (Race Across the West) | 8,848 m of climbing in 6 h 41 min |
| Carbs per hour | 58 g average | About 149 g |
| Mean glucose while riding | 91 mg/dL | 160 mg/dL |
| Time below 70 mg/dL | 9.15% | 0% |
| Time above 140 mg/dL | 1.77% | 100% |
| Result | Course record | Finished the 6 h 41 min attempt |
One rider spent nearly a tenth of his time under 70 mg/dL and set a course record. The other sat above 140 mg/dL the whole day and also performed. In short: there’s no single “good” glucose line for racing.
The authors say readings outside 70 to 140 mg/dL are common in athletes and likely adaptive. Both athletes also show why a glucose curve can’t set your carbs per hour. Intake changes the curve. The curve can’t tell you what intake you need.
What Eating 58 g vs 149 g an Hour Teaches
Look at the intake gap first. The relay rider averaged 58 g of carbs an hour. That number hid a slide: about 92 g an hour on day one, falling to 31 g an hour by day three. His glucose ranged from 54 to 169 mg/dL. He still set the course record.
The Everesting rider ate about 149 g an hour, close to 1,000 g in total. His glucose stayed between 151 and 178 mg/dL. That’s a very flat line. In short: lots of carbs gave a high, steady curve, and fewer carbs gave a lower, bumpier one.
Neither curve was a verdict. Each is one athlete on one day, so you can’t copy either rider’s numbers. What carries over is the method. Change your intake, watch the curve, and judge it against how you felt and how you performed.
The UCI Rule: What’s Banned and What’s Not
If you race on the road, know this. UCI Article 1.3.006 bans devices that capture glucose or lactate during competition. Kristen Faulkner was disqualified from the 2023 Strade Bianche for wearing a sensor she said wasn’t recording.
The ban covers competition only. Training use is allowed. Riders with type 1 diabetes are exempt. Check the current UCI rulebook before race day.
Your Options Now
Supersapiens, the platform behind most athlete CGM data, closed all memberships on March 5, 2024. Its sensor came from Abbott. These are the main options today.
| Option | Who it’s for | Price | Built for exercise accuracy? |
|---|---|---|---|
| Abbott Lingo | Adults without insulin, OTC (FDA cleared June 2024) | UK: £89 per sensor | No |
| Abbott Libre | Same sensor family as Supersapiens | Varies by region | Not validated for sport |
| Dexcom Stelo | Adults 18+ not on insulin, OTC (FDA cleared March 2024) | From $89.99 for two sensors | No |
Neither Lingo nor Stelo is validated for athletic exercise accuracy. Check current prices with each maker.
How to Read a CGM Curve on a Long Session
Raw lines are easy to misread. Use the same steps every time.
- Log your feeds. Write down the clock time of every gel, drink and chew. Without this, the curve is a squiggle.
- Set a baseline. Note your average over the first 20 minutes, before the first feed.
- Wait out the lag. After a gel, give the sensor 12 to 24 minutes before you judge the response. The reading you see is old news.
- Match it to your body. Did the dip line up with heavy legs, or with nothing at all?
- Repeat. Compare across three or four long sessions before you change anything.
Lag matters most at the end of a hard interval. Say you finish a hard climb and the screen shows a falling number. Your blood sugar may already be climbing back up, because you’ve eased off and taken a gel. If you panic and eat more, you can stack extra carbs on a gut that’s already full. Wait, then look again.
Use It as a Check, Not a Prescription
Start with the plan, then test it. Set a carb target first. The AthleteOS fueling calculator turns your body size, race length and gut-training history into a carbs-per-hour number. Then wear a CGM on a long run or long ride and see how that plan behaves. Note that AthleteOS doesn’t read CGM data. You compare the two yourself.
Look for patterns, not moments:
- Repeated dips under 70 mg/dL across several long sessions, with fatigue.
- A steep fall right after a big gel, then a bounce.
- Readings that match how your legs and stomach feel.
One dip is noise. Three of them are a pattern worth testing.
Here’s a made-up example, using illustrative numbers. Dana is 41 and trains for a 3:40 marathon. Her plan says 60 g of carbs per hour. On two 2-hour long runs, her sensor dipped under 70 mg/dL near the 1:40 mark. She also felt flat at the same point both days. She raised intake to 75 g per hour over three weeks, building stomach tolerance on easy runs. The next long run had no dip, and she felt even.
Dana didn’t follow the sensor. She followed the sensor plus her legs, and she changed intake slowly. If you’re building long-run volume, read how Zone 2 builds the aerobic base so your easy days stay easy while you test fuel.
When a CGM Isn’t Worth Buying
Skip it if you haven’t dialed in a basic plan yet. Eating 60 to 90 g per hour in training and tracking gut comfort costs nothing. Start with our guides to finding your personal carb absorption ceiling and training your gut to take more carbs. Buy a sensor only when it answers a question you actually have.
A CGM makes sense if you’re curious, you’ve already got a working plan, and you’ll test it for a few weeks. It makes less sense as a shortcut.
Your sensor shows a cup. Your plan feeds the tank.
Ready to set your target? Create your AthleteOS account and build your carb plan.