As incentive for engaging in physical activity, outcome expectancy represents a central construct of social cognitive theories that have been extensively utilized in physical activity promotion (Williams et al., 2005). In most existing studies, outcome expectancy has been measured using the multiplicative combination of outcome likelihood (probability that a certain action will lead to a certain outcome) and outcome values (values the individual assigns to the possible outcomes). However, the reliability of this calculation method has been questioned because the amount of variance accounted for in outcome expectancy by the multiplicative composite score may vary depending on the numerical scale used to measure outcome likelihood or outcome values (Gagne & Godin, 2000). To our knowledge, few studies have been conducted to investigate the temporal stability of outcome likelihood, outcome values, and outcome expectancy in physical education settings using longitudinal research designs. Therefore, the purpose of this study was to examine the temporal stability of these variables in middle-school physical education over the course of one year. The participants were 101 sixth graders and 99 seventh graders (109 boys, 91 girls) in a southeastern state. Self-report scales were used to assess outcome likelihood and outcome values in physical education (Goudas & Dermtizaki, 2004) over a three-semester period that started in spring 2006. Outcome expectancy scores were generated through the cross-products of outcome likelihood and outcome values. The internal reliabilities of these variables were calculated using Cronbach's alpha coefficients over the three semesters. The respective reliability coefficients were as follows: .90, .93, and .91 for outcome likelihood; .93, .95, and .94 for outcome values; and .92, .96, and .95 for outcome expectancy. The findings indicated acceptable and consistent reliability coefficients of outcome likelihood, outcome values, and outcome expectancy over time. Therefore, these scales can accurately capture students' motivation toward active participation in physical education. Second, the results revealed an increase of the reliability coefficients over one school year. This finding may reflect that as students mature they become increasingly aware of the benefits of physical activity, and they can accurately report their perceptions concerning the potential outcomes and values of physical education. These findings also suggest using the cross-products of outcome likelihood and outcome value to measure outcome expectancy is sound. Future health promotion researchers would need to longitudinally measure outcome expectancy and physical activity changes to capture the temporal changes in physical activity motivation and behavior.