The aim of this study was to determine the type and the number of accelerometer monitoring days needed to predict weekly sedentary behaviour and physical activity in obese youth. Fifty‐three obese youth wore a triaxial accelerometer for 7 days to measure physical activity in free‐living conditions. Analyses of variance for repeated measures, Intraclass coefficient (ICC) and regression linear analyses were used. Obese youth spent significantly less time in physical activity on weekends or free days compared with school days. ICC analyses indicated a minimum of 2 days is needed to estimate physical activity behaviour. ICC were 0·80 between weekly physical activity and weekdays and 0·92 between physical activity and weekend days. The model has to include a weekday and a weekend day. Using any combination of one weekday and one weekend day, the percentage of variance explained is >90%. Results indicate that 2 days of monitoring are needed to estimate the weekly physical activity behaviour in obese youth with an accelerometer. Our results also showed the importance of taking into consideration school day versus free day and weekday versus weekend day in assessing physical activity in obese youth.