Twelve bicycle and Peloton sessions were selected. The starting value is the continuous reading nearest to the exercise note. Glucose then had to fall by at least 5 mg/dl within 90 minutes and rebound by at least 15 mg/dl no later than 180 minutes after exercise started. Documented carbohydrate in the preceding two hours and an immediately preceding additional ride led to exclusion.
| Date | Session | Start | Nadir | later peak | Drop | Bounce |
|---|---|---|---|---|---|---|
| 2024-01-09 | Bicycle | 125 | 117 | 142 | −8 | +25 |
| 2026-01-04 | Peloton beginner | 156 | 123 | 142 | −33 | +19 |
| 2026-01-19 | Bike training | 158 | 145 | 181 | −13 | +36 |
| 2026-01-21 | Bicycle | 136 | 110 | 153 | −26 | +43 |
| 2026-03-30 | Low impact | 155 | 128 | 149 | −27 | +21 |
| 2026-04-02 | Low impact | 156 | 135 | 158 | −21 | +23 |
| 2026-04-08 | Cooldown + low impact | 145 | 133 | 176 | −12 | +43 |
| 2026-04-13 | Low impact | 148 | 138 | 159 | −10 | +21 |
| 2026-06-14 | 60 min rock | 116 | 107 | 130 | −9 | +23 |
| 2026-06-23 | Low impact | 124 | 115 | 136 | −9 | +21 |
| 2026-07-05 | 90s pop + low impact + cooldown | 108 | 99 | 115 | −9 | +16 |
| 2026-07-07 | German rock up to 400 W | 152 | 129 | 144 | −23 | +15 |
My observation: the curve falls, then returns
During harder or longer rides, muscle uptake can push the CGM reading down first. The curve may then rise again. The pattern also appeared in some sessions labelled Low Impact, so intensity alone cannot be identified as the trigger. In these twelve selected traces the rebound ranged from 15 to 43 mg/dl. The later peak did not always exceed the starting value.
Why liver glucose output is plausible
Exercise requires a supply of glucose to working muscle. At high intensity, counter-regulatory hormones can substantially increase hepatic glucose output. Published physiology describes situations in which liver glucose production exceeds utilisation during intense exercise. That makes a bounce plausible. My CGM data do not measure liver flux or hormones and therefore do not prove the mechanism.
Other explanations can overlap
- delayed absorption from an earlier meal,
- the lag between blood and interstitial CGM values,
- the dawn phenomenon during morning sessions,
- changing intensity within a ride,
- recovery periods and declining muscle uptake.
What changes for me
I no longer judge a ride by the lowest value immediately afterwards. The next two to three hours matter. After a steep fall I reduce intensity, keep observing and avoid rushed countermeasures. People using insulin or glucose-lowering medication need an individually agreed plan.
Method and limits
The selection rule was applied retrospectively to existing notes. Intensity was not consistently standardised by power, heart rate or zone. The description ‘hard’ is directly supported only where notes mention, for example, up to 400 watts or a long rock/pop block. Undocumented meals, time of day and CGM lag may affect the curves. This is a structured personal case series, not a frequency estimate for all rides.
Sources and methodological context
- American Diabetes Association, Standards of Care in Diabetes—2026: Physical activity and exercise
- Trefts et al., Exercise and the Regulation of Hepatic Metabolism
- Bergenstal et al., Glucose Management Indicator (GMI)
- Zaharieva et al., CGM lag during aerobic exercise
- Emhoff et al., Gluconeogenesis and hepatic glycogenolysis during exercise
The values in this article come from my own sensor (FreeStyle Libre 3) and are a single-case observation (n = 1) – they show a pattern in me, not a rule for you. Every statement is labelled as established, plausible or open, and proximity in time never counts as proof of a cause. How I handle time windows, sensor lag, influencing factors and approval is collected on one page. Read the methodology