Early school times fundamentally clash with the late sleep of teenagers. This mismatch results in chronic sleep deprivation posing acute and long-term health risks and impairing students' learning. Despite immediate short-term benefits for sleep, the long-term effects of later starts remain unresolved. In a pre-post design over 1 year, we studied a unique flexible school start system, in which 10–12th grade students chose daily between an 8:00 or 8:50AM-start. Missed study time (8:00–8:50) was compensated for during gap periods or after classes. Based on 2 waves (6–9 weeks of sleep diary each), we found that students maintained their ~ 1-h-sleep gain on later days, longitudinally (n = 28) and cross-sectionally (n = 79). This gain was independent of chronotype and frequency of later starts but attenuated for boys after 1 year. Students showed persistently better sleep quality and reduced alarm-driven waking and reported psychological benefits (n = 93) like improved motivation, concentration, and study quality on later days. Nonetheless, students chose later starts only infrequently (median 2 days/week), precluding detectable sleep extensions in the flexible system overall. Reasons for not choosing late starts were the need to make up lost study time, preference for extra study time and transport issues. Whether flexible systems constitute an appealing alternative to fixed delays given possible circadian and psychological advantages warrants further investigation.