The generality of a multilevel factorial model of social competence (SC) for preschool children was tested in a 5-group, multinational sample (N = 1,540) using confirmatory factor analysis. The model fits the observed data well, and tests constraining paths for measured variables to their respective first-order factors across samples also fit well. Equivalence of measurement models was found at sample and sex within-sample levels but not for age within sample. In 2 groups, teachers' ratings were examined as correlates of SC indicators. Composites of SC indicators were significantly associated with both positive and negative child attributes from the teachers' ratings. The findings contribute to understanding of both methodological and substantive issues concerning SC in young children.