This study examined longitudinal relations between sleep and body mass index (BMI) from late childhood ([X-bar] age = 9.44 at T1) to early adolescence ([X-bar] age = 11.36 at T3) with a relatively large (n = 273 at T1) and diverse (35% African Americans) sample. Sleep was assessed with actigraphy-based sleep minutes and self-reported sleep problems. Latent growth modeling revealed associations between sleep and BMI that varied based on the sleep parameter and sex of the child. In particular, although boys reported more sleep problems and had higher BMI than girls across time, self-reported sleep problems at age 9 predicted higher BMI at age 11 among girls (but not boys). Furthermore, more minutes of actigraph-assessed sleep at age 9 only predicted less growth in BMI among the girls. Possible mechanisms and intervention implications are discussed.