Purpose: Background: Physical activity is a complex behavior which involves the interaction of multilevel factors at the individual, social and environmental level. However, previous studies have largely focused on psychological and/or social environmental factors and the direct impact of such factors on physical activity. There are few studies having examined how multilevel factors may interact to influence activity level. Therefore, the purpose of the present study was to examine both direct and indirect effects of multilevel factors on school-based physical activity in Japanese adolescent boys. Methods: In this cross-sectional survey of the Japanese adolescent lifestyles, 379 junior high school boys were invited to complete self-report measures of age, grade, weight, height, self-efficacy, social support (family, friends and teachers), school physical environment (equipment, facilities and safety) and average minutes per week of physical activity during lunch time and after-school hours occurring at school. Structural equation modeling analyses controlling for age were utilized to examine the effects of body mass index (BMI), self-efficacy, social support and school physical environmental variables on lunchtime and after-school physical activity. Results: During lunch time, self-efficacy exhibited direct positive effects on physical activity. BMI, facilities, and safety were indirectly associated with lunchtime physical activity through self-efficacy. However, there were no significant relationships of equipment and social support with lunchtime physical activity. During after-school hours, family support and facilities directly affected physical activity. Self-efficacy was indirectly related with physical activity through family support. BMI, equipment, and safety indirectly affected physical activity through self-efficacy and/or family support. Conclusion: Effects of multilevel factor on physical activity among adolescent boys differed according to context, which implies that interventions to promote physical activity should be context-specific. Findings encourage the development of future effective interventions to promote physical activity through self-efficacy during lunch time as well as family support during after-school hours.