This paper investigates the frequency and the duration of participation in physical activity by children who go to school in the city of Dhaka, Bangladesh. The study uses data from a 2012 activity-based travel diary survey of 245 students in Grades 6 to 10 in the Dhaka metropolitan area. To explore the trade-offs between factors that affect the frequency of out-of-home physical activity, including the use of active transportation, this study used a zero-inflated Poisson (ZIP) regression model. Additionally, parametric hazard models were estimated to examine the duration of physical activities. The results of the ZIP count model suggest that personal and household characteristics, attributes of the built environment, and time commitments for other mandatory and discretionary activities influence the frequency of participation in physical activity. For example, an increase in the number of children in the household and the presence of sidewalks increased the frequency of out-of home, out-of-school physical activity if everything else was held equal. In contrast, travel time to school, the presence of traffic intersections, and the time commitment for tutoring sessions negatively affected participation in physical activity. In the modeling of the duration of out-of-home physical activity, this study found that a Weibull parametric hazard model outperformed a log logistic model. The duration was influenced by sociodemographic characteristics, spatial context, and escort arrangements. For instance, car ownership and a parental escort increased the duration of physical activity. The paper offers an in-depth behavioral understanding of children's physical activities in a developing country.