Among 209 naturally cycling women without diabetes (mean age 43, BMI 24.3) wearing continuous glucose monitors for up to 14 days, average glucose and glucose management indicator did not change across the cycle. Six variability metrics did, bottoming out at 38-41% of cycle progress and peaking at 80-85%, in the late luteal phase, with swings of 6-10% of the cycle mean. Matched men assigned pseudo-cycles showed no pattern. Luteal breakfasts produced 15-23% larger postprandial excursions, and the effect held in composition-matched breakfasts eaten by the same women.
The finding fits what clinicians already see in type 1 diabetes, where insulin needs rise in the luteal phase, but it is rarely documented in metabolically healthy women. The key point for longevity is that these are modest fluctuations around an unchanged mean, so they are not evidence of harm and should not alarm anyone. The practical lesson is interpretive: a single CGM reading, an oral glucose test or a wearable-guided diet experiment may read differently depending on cycle timing. Limitations matter. Ovulation was estimated rather than hormonally confirmed, the monitoring window covers roughly one cycle per woman, and the sample is mostly premenopausal women in their forties. This is a preprint that has not been through peer review, so effect sizes may shift. Overall the work is confirmatory and incremental, but a useful corrective for anyone interpreting CGM data.