PURPOSE: The psychometric properties of the Self-Motivation Inventory for children (SMI-C) were established using tests of factorial validity, factorial invariance, latent mean structure, and predictive validity. METHODS: Adolescent girls from two cohorts (N = 955 and 1797) completed the SMI-C and measures of physical activity, team sport involvement, cardiorespiratory endurance, and body fatness in the 8th grade; participants in cohort 2 ( N= 1658) also completed the measures 1 yr later in the 9th grade. The data were analyzed with exploratory and confirmatory (CFA) factor analysis and structural equation modeling (SEM). RESULTS: The 20-item SMI-C was best represented by a single substantive factor, but there were method effects among the negatively worded items. CFA indicated that a single-factor model fit the nine positively worded items and exhibited strong evidence of cross-validity and factorial invariance between races and across 1 yr; there were no differences in latent means between black and white girls. SEM indicated that the nine-item SMI-C had direct effects on moderate (gamma(11) = 0.16) and vigorous (gamma(21) = 0.22) physical activity and team sport involvement (gamma(31) = 0.29); the effects were invariant between black and white girls, independent of cardiorespiratory endurance and body fatness, and consistent across a 1-yr period. CONCLUSION: The positively worded nine-item version of the SMI-C can be used in cross-sectional, prospective cohort, and intervention studies that examine self-motivation as a putative moderator or mediator of physical activity among black and white adolescent girls.