Objective To investigate the longitudinal changes in children's recess and lunchtime physical activity levels and in the contribution of recess and lunchtime to daily physical activity levels over 5 years among 5–6- and 10–12-year olds.

Methods Data were drawn from two longitudinal studies that were conducted in metropolitan Melbourne, Australia. Boys and girls (n=2782) aged 5–6 years and 10–12 years participated in baseline (T0) measures. Physical activity (n=2490) was measured every 60 s for eight consecutive days using hip-mounted accelerometry. Subsequent measurements were taken at 3-year (T1; n=773) and 5-year (T2; n=634) follow-up. Physical activity intensities were derived using age-adjusted cut-points. Sedentary time was defined as 100 counts/min. Longitudinal data were analysed using three-level (time, child, school) multilevel analyses, stratified by sex and cohort, and adjusted for potential confounding variables.

Results Significant decreases in recess and lunchtime moderate and vigorous physical activity were observed (p<0.001), with larger decreases occurring in the older cohort. Associated increases were observed in sedentary time over time (p<0.01). Although the contribution of recess to daily moderate intensity physical activity increased in the younger cohort over time (p<0.001), significant decreases were observed in the older cohort (p<0.001).

Conclusion Physical activity levels during recess and lunchtime decreased in both cohorts over time. Decreases in the contribution of recess and lunchtime to older children's daily physical activity were also observed. Interventions are needed in both primary and secondary schools to promote physical activity levels during recess and lunchtime, particularly during the early years of secondary school.