Free time on the playground during the school day is perhaps the most universal opportunity for children in elementary school to be vigorously active in a social setting. Of concern, therefore, are research findings that some children do not take an active part in free time on the playground (Bouffard, Watkinson, Thompson, Causgrove Dunn & Romanow, 1996; Smyth & Anderson, 2000). The purpose of this study was to examine the extent to which children's playground activity choices could be predicted by Eccles and colleagues' expectancy-value model of achievement motivation (Eccles et al., 1983; Eccles, Wigfield & Schiefele, 1998; Wigfield & Eccles, 1992). The model predicts that individuals' decisions to engage are directly linked to their expectations for successful performance, and the values they hold for the various options perceived to be available. Study participants included 242 children in Grades 1 to 4 (99 boys, 143 girls) from seven schools. Activities engaged in were assessed using the Activities for Daily Living in Physical Play (ADL-PP) (Watkinson et al., 2001), a self-report form completed immediately following recess. Task-specific expectancies (perceptions of competence) and values (interest and importance) were assessed using the playground activities contained in the ADL-PP. Multivariate prediction analysis of cross classification of qualitative variables (Hildebrand, Laing, & Rosenthal, 1977) was used to test a priori predictive statements about children's recess activity choices based on Eccles et al's theory. The results strongly supported the predictions from the model (<font face="Symbol">D=.676, p < .0001), indicating that knowledge of the state of the independent variables, expectancies and values, resulted in a 68% reduction in error predicting engagement. Moreover, partial <font face="Symbol">D's suggested importance contributed significantly to the prediction of playground activity engagement, over and above the predictive value of perceived competence (partial <font face="Symbol">D=.629, p < .0001). In contrast, perceived competence added, on average, no predictive value beyond that achieved by knowledge of importance (partial <font face="Symbol">D=.0, p=.50). This is consistent with previous research findings from other domains that suggest subjective task values are more influential in initial decisions to engage in an activity, while expectancies are more influential in persistence, effort, and performance (Eccles et al., 1998; Wigfield & Eccles, 1992). Overall, the results strongly support the application of Eccles' expectancy-value model to the playground setting, and increase our understanding of the factors affecting children's decisions to take part in activities on the playground.Keyword(s): elementary education, physical activity, play