Self-efficacy theory (Bandura, 1997) and achievement goal theory (Nicholls, 1989) have proven fruitful in explaining students' motivation and achievement behaviors in physical education. In support of achievement goal theory, dispositional achievement goals and perceptions of the motivational climate have been found to interact and impact patterns of motivated behaviors.  Furthermore, research has repeatedly shown that self-efficacy is a critical antecedent to behavior. It is possible that students' achievement goals and the motivational climate influence self-efficacy, which in turn, influences achievement behaviors (Cumming & Hall, 2004; Kavussanu & Roberts, 1996). The mediating role of self-efficacy, however, has not been investigated with the variables of the 2 x 2 dispositional achievement goal model (mastery-approach [MAp], mastery-avoidance [MAv], performance-approach [PAp], and performance-avoidance [PAv]) in physical education. The purpose of this study, therefore, was to examine the mediating effect of self-efficacy on relationships among middle school students' achievement goals and perceptions of the motivational climate and physical activity (PA) levels in physical education class. The four dispositional achievement goals, perceptions of the motivational climate (task- and ego-involved climates), and self-efficacy were measured in a sample of 225 students (112 boys, 113 girls) in a public school and were used to predict students' PA levels. PA levels were measured with Actical activity monitors (Mini-Mitter Co., Inc., Bend, OR) for one regular physical education class. Four multiple regression analyses were employed to test the three criteria for examining the presence and significance of a mediation effect (Baron & Kenny, 1986). The results suggested that the link between achievement motivation and self-efficacy was significant. Specifically, MAp (b = .35, p < .01) and task-involved climates (b = .25, p < .01) emerged as significantly positive predictors of self-efficacy; Second, MAp (b = .18, p < .05) and task- involved climates (b = .17, p < .05) significantly affected PA levels. As well, self-efficacy (b = .49, p < .01) also significantly affected PA levels; Third, MAp (b = .02, p = .86) and task- involved climates (b = .06, p = .47) failed to exert significant effects on PA levels when the effect of self-efficacy was controlled. The findings supported the mediating effect of self-efficacy. Thus mastery based achievement goals and motivational climates foster PA because the students feel more efficacious or better able to move with confidence. This is encouraging as it provides insight into how teachers might positively affect students' motivation and PA levels.