Objective Children's physical activity levels are difficult to establish on a day-to-day or season-to-season basis. Most studies have attempted to measure reliability in habitual settings. This study investigated the variability in children's physical activity during recess. Methods Fifteen boys and 19 girls (aged 6 to 11 years) from 2 schools in North West England wore heart rate monitors for 5 consecutive days in summer and winter terms to assess day-to-day and seasonal variability during school recess. Data were collected in 2004. Repeated measures ANOVA's and intraclass correlations (ICC) analysed the day-to-day and seasonal variability in children's moderate-to-vigorous (MVPA) and vigorous physical activity (VPA) data. Results There were no significant differences in children's MVPA and VPA across days and seasons. ICCs for MVPA across 2 days ranged from 0.75 to 0.85 in summer, and from 0.53 to 0.81 in winter. Three-day MVPA ICCs were 0.83 in summer and 0.71 in winter. Conclusions The results revealed no significant variation in children's recess physical activity levels across days and seasons. Whilst children were free to choose their recess activities in school, the results suggested that children were relatively consistent in their choices, limiting physical activity variability.