Background. This study evaluated the effects of the Lifestyle Education for Activity Program (LEAP), a comprehensive school-based intervention emphasizing changes in instruction and school environment, on variables derived from social-cognitive theory (SCT) as mediators of change in physical activity among black and white adolescent girls. Methods. Twenty-four high schools paired on enrollment size, racial composition, urban, suburban, or rural location, and class structure were randomized into control (n = 12) or experimental (n = 12) groups. There were 1038 girls in the control group and 1049 girls in the experimental group. The multicomponent intervention emphasized the enhancement of self-efficacy and development of behavioral skills by using curricular activities within physical education classes and health education instruction. The primary outcomes were self-efficacy, outcome-expectancy value, goal setting, satisfaction, and physical activity. Results. Latent variable structural equation modeling indicated that: (1) self-efficacy and satisfaction exhibited synchronous, cross-sectional relationships with physical activity; (2) the intervention had direct effects on self-efficacy, goal setting, and physical activity; and (3) self-efficacy partially mediated the effect of intervention on physical activity. Conclusions. To our knowledge, this study provides the first evidence from a randomized controlled trial that manipulation of self-efficacy results in increased physical activity among black and white adolescent girls. The results encourage the use of self-efficacy as a targeted, mediator variable in interventions designed to increase physical activity among girls. © 2004 The Institute For Cancer Prevention and Elsevier Inc. All rights reserved.