Physical activity declines during childhood and determinants of this decline are poorly understood. The purpose of this prospective study was to examine modifiable psychological and social correlates of physical activity change over 20 months.

370 girls and 362 boys in public schools, studied through the 4th and 5th grades.

A physical activity index was computed from child reports, parent reports, and objective activity monitoring. Predictor variables were assessed by child and parent surveys.

In multiple regressions, after adjustment for demographic variables, psychological variables explained 4% of the variance in boys’ physical activity change, parent variables explained 8%, and interactions with skinfolds explained 3%, for a total of 15% (unadjusted). Regarding change in girls’ physical activity, psychological variables explained 3%, parent variables explained 1.5%, and interactions with skinfold explained 1.5%, for a total of 6% explained variance (unadjusted). There were differences in predictors among children with high and low levels of body fat.

Children’s preferences for physical activity and frequency of parents transporting children to activity locations explained significant proportions of variance for girls and boys. The results support a dynamic model of the determinants of children’s physical activity, but most of the variance was left unexplained.