This essay investigates the impacts of primary care giver’s (PCG) time allocation and food expenditure choices on childhood obesity using national panel study of income dynamic (PSID) data. A triangular system of equations is derived and estimated under parametric and semi-parametric model settings. The performances of the two modeling strategies are compared using predictive ability measures with the aid of bootstrap method. Test results suggest relatively better performance of the semi-parametric model than parametric model. Nevertheless, the comparison of the estimates from both parametric and semi-parametric estimation indicates no dramatic changes in our findings. Our results do not suggest significant impacts of PCG’s labor force participation choices, involvement in children’s outdoor activity, and household food expenditures on children’s Body Mass Index (BMI). However, the estimates from both iterated seemingly unrelated regression (SUR) and semi-parametric polynomial estimation indicate that parents’ BMI significantly influence children’s BMI. Obese parents tend to have obese children. Furthermore, physical activity appears to have weak correlation with children’s BMI. More physical activity time does not necessarily lead to lower BMI of children.