OBJECTIVES: Describe (1) time-segment specific changes in physical activity (PA) into adolescence, (2) differences in change in PA between specific time-segments (weekdays-weekends, in-school-out-of-school, out-of-school-weekends, lesson-time-lunch-time), and (3) associations of change in time-segment specific with overall PA. DESIGN: Longitudinal observational study (4-year follow-up). METHODS: Children from the SPEEDY study (n=769, 42% boys) had PA measured by accelerometer for at least three days at ages 10.2±0.3, 11.2±0.3 and 14.3±0.3years. Changes in moderate-to-vigorous PA (ΔMVPA, minutes ≥2000counts/minute [cpm]) and total PA (ΔTPA, average cpm) during weekdays, weekends, in-school, out-of-school, lesson-times and lunch-times, were tested using three level (age, individual, school) mixed-effects linear regression. Differences in ΔMVPA/ΔTPA between time-segments were tested using time-segment×age interaction terms. Associations of four-year time-segment specific ΔMVPA/ΔTPA with four-year overall ΔMVPA/ΔTPA were tested using two level (time-segment specific ΔMVPA/ΔTPA, school) mixed-effects linear regression. RESULTS: MVPA and TPA declined in all time-segments, except lesson-time MVPA. Annual ΔMVPA and, for boys only, ΔTPA was greater on weekends than weekdays (beta±SE for interaction term: boys, -3.53±0.83min, -29.64±7.64cpm; girls, -2.20±0.64min) and out-of-school (boys, -4.36±0.79min, -19.36±8.46cpm; girls, -2.44±0.63min). ΔMVPA and ΔTPA during lunch-time was greater than during lesson-time (boys, -0.96±0.20min, -36.43±6.55cpm; girls, -0.90±0.13min, -38.72±4.40cpm). ΔTPA was greater out-of-school than in-school (boys, -19.89±6.71cpm; girls, -18.46±6.51cpm). For all time-segments, four-year ΔMVPA/ΔTPA was positively associated with four-year overall ΔMVPA/ΔTPA (all p<0.042), except for girl's in-school and lunch-time TPA. CONCLUSIONS: Interventions focused on PA maintenance could target all time-segments, but weekends and out-of-school may be particularly advantageous due to the relatively large declines observed.