Objectives To determine in children aged 0–7 years (1) cross-sectional relationships between body mass index (BMI) and sleep duration, and whether (2) sleep duration predicts later BMI and/or (3) BMI predicts later sleep duration.

Design Longitudinal Study of Australian Children, Waves 1 and 2.

Participants Infants aged 0–1 years (Wave 1), followed at age 2–3 years (Wave 2); children aged 4–5 years (Wave 1), followed at age 6–7 years (Wave 2).

Main outcome measures BMI, sleep duration by time-use diary, parent-reported sleep problems (none/mild vs moderate/severe). Analyses: Cross-sectional, ANOVA (sleep duration) and χ² (sleep problems); longitudinal, linear regression.

Results 3857 (76%) infants and 3844 (77%) children had BMI and sleep data. At every wave, approximately 15% and 5% of children were overweight and obese, respectively. Obesity was not associated with sleep duration at 0–1, 2–3 or 4–5 years, though obese 6–7-year olds slept approximately 30 min less (p<0.001). Sleep problems were similar across BMI categories at all ages. Wave 1 sleep duration did not predict Wave 2 BMI, nor did Wave 1 BMI predict Wave 2 sleep duration.

Conclusions In these large child population cohorts, sleep duration did not predict obesity up to age 6–7 years. Current trials of sleep interventions to prevent or manage obesity in young children may be premature.