To prospectively examine the association between device-measured sleep regularity and incidence of type 2 diabetes (T2D) in a population-based sample of adults. We also examined if meeting sleep duration recommendations attenuated or eliminated the effects of irregular sleep on T2D. We conducted a prospective cohort study of adults aged 40-79 years participating in the UK Biobank accelerometer substudy. Participants wore wrist-attached accelerometers for a duration of 7 days, which was used to compute the Sleep Regularity Index (SRI). Participants were categorized as irregular (SRI <71.6), moderately irregular (SRI between 71.6 and 87.3), and regular (SRI >87.3) sleepers. T2D diagnosis was obtained through self-reports and health records. We analyzed data from 73,630 individuals observed for 8 years, without a history of T2D and without an event in the first year of follow-up. Compared with regular sleepers, irregular (hazard ratio [HR] 1.38; 95% CI 1.20-1.59) and moderately irregular sleepers (HR 1.35; 95% CI 1.19-1.53) were at higher risk of T2D incidence. Dose-response analyses treating SRI as a continuous measure showed higher T2D incidence with SRI scores <80. Meeting current sleep duration recommendations did not counteract the adverse effects of irregular (HR 1.35; 95% CI 1.09-1.66) or moderately irregular (HR 1.29; 95% CI 1.08-1.54) sleep on T2D incidence. Moderate and high sleep irregularity were deleteriously associated with T2D risk, even in participants who slept ≥7 h per night. Future sleep interventions will need to pay more attention to consistency in bedtimes and wake-up times, in addition to sleep duration and quality.