Objective: To explore the effects of an innovative school-based intervention for increasing physical activity. Methods: 226 children (5-7. years old) randomly selected from 12 Australian primary schools were recruited to a cluster randomised trial with schools randomly allocated to intervention or control conditions. The 13-week intervention comprised: (1) altering the school playground by introducing loose materials and (2) a teacher-parent intervention exploring perceptions of risk associated with children's free play. The primary outcomes were total accelerometer counts and moderate-vigorous physical activity during break times. Testing took place in Sydney, 2009-2010. Results: 221 participants were tested at baseline. Mixed-effect multilevel regression revealed a small but significant increase from the intervention on total counts (9400 counts, 95% CI 3.5. -. 15.2, p. =. 0.002) and minutes of MVPA (1.8. min, 95% CI 0.5-3.1, p. =. 0.006); and a decrease in sedentary activity (2.1. min, 95% CI 0.5-3.8, p. =. 0.01) during break times. We retested children in one intervention school after 2. years; they maintained the gains. Conclusions: Capturing children's intrinsic motivations to play while simultaneously helping adults reconsider views of free play as risky provided increases in physical activity during break times. Using accelerometry as the sole measure of physical activity may underestimate the effect.Trial registration: ACTRN12611000089932.