Background: Prevalence of childhood overweight and obesity has risen dramatically over the last few decades, suggesting a causative role for environmental factors, including activity, diet and eating behaviours.Objectives: To determine the prevalence, change in prevalence and degree of tracking of overweight and obesity in a birth cohort of Dunedin children from age three to seven years and to simultaneously examine multiple putative risk factors and potential confounders for increased BMI in childhood, focusing on those potentially amenable to intervention.Methods: This thesis describes a prospective observational study, which recruited 241 children at age three (baseline data) and followed them to age seven (longitudinal data). Body Mass Index (BMI), fat mass and fat free mass were measured six-monthly with bioelectrical impedance and at ages five and seven using dual energy x-ray absorptiometry. Risk factors assessed included physical activity and sleep duration measured annually with accelerometry; dietary intake, physical inactivity including screen time, and family factors (parental BMI, socio-demographic factors, birth weight (BW) and smoking during pregnancy) measured by questionnaire. Main outcome measures were prevalence of overweight and obesity (1) at each age and multivariate models developed to determine associations with BMI cross-sectionally at baseline and with BMI and body composition longitudinally to age seven.Results: At baseline, combined prevalence of overweight and obesity was 26.1%. In multivariate analyses Maori ethnicity, antenatal smoking, higher BW and higher paternal BMI were cross-sectionally associated with increased BMI. Higher maternal pressure to eat was associated with lower BMI in this analysis. An interaction effect existed between maternal smoking and BW. Whereas in the children of non-smokers BW was positively associated with subsequent BMI, this relationship was reversed for children of smokers. Longitudinal follow-up revealed combined prevalence decreased to 18.1% at seven although BMI tracked strongly. Overweight and obesity was more prevalent in girls from age five, and girls had lower fat free mass and higher fat mass at all ages. Multivariate analysis identified two potentially modifiable risk factors longitudinally associated with BMI. Antenatal smoking remained significant with a difference in BMI of +0.96 kg/m2 (95% confidence interval: 0.33 to 1.58). Each additional hour of sleep at ages three to five was associated with a reduction in BMI of 0.48kg/m2 (0.01 to 0.95) and a reduced risk of overweight of 0.39 (0.24 to 0.63) at seven. This association persisted after adjustment for earlier BMI and was explained by greater differences in fat mass index (-0.43, -0.82 to -0.03) than in fat free mass index (-0.21, -0.41 to -0.00). Other independent predictors of increased BMI were female sex, Maori ethnicity, high maternal BMI and high BW but the negative cross-sectional association with maternal pressure to eat demonstrated at three did not persist in longitudinal analyses adjusting for earlier BMI. Conclusions: Young children exposed to tobacco in-utero and who sleep less are at increased risk of becoming overweight, even after adjustment for multiple confounding factors. Maternal pressure to eat may be the result of concerns regarding perceived poor growth, rather than a protective factor in the development of overweight.