Social-emotional competence (SEC), or effectiveness of social interaction, plays a central role in children’s health and well-being. The three goals of the current study were to describe the development of SEC during a preschool year; identify an appropriate factor structure for observed teacher-child interactions; and predict SEC growth from children’s characteristics (emotional competence, language risk, gender, and race/ethnicity), teacher-child relationship quality, and classroom characteristics (relational climate, observed teacher-child interaction quality, and curriculum). Children’s social competence, anger/aggression, effortful control, and emotion knowledge (N =822) was assessed at three time points during a preschool year via teachers’ reports and behavioral assessments. In the fall, teachers reported the quality of their relationships with children and teacher-child interactions were observed in classrooms. Aim 1 results indicated that children exhibited linear increases in effortful control and social competence and stability in anger/aggression, although social competence was the only construct where linear change varied among children. Due to a lack of longitudinal measurement invariance, growth in latent emotion knowledge could not be evaluated. Several gender and racial/ethnic differences were identified in SEC intercepts, but not the social competence slope. Language risk and impulsivity were consistent predictors of SEC intercepts. Aim 2 results indicated that teacher-child interaction quality was primarily unidimensional. Finally, results from aim 3 indicated that children’s emotional competence at the beginning of the year and classroom relational climate were predictive of growth in social competence. End-of-year social competence levels were associated with supportive teacher-child relationship quality (particularly among girls), high emotional competence, low language risk, and supportive classroom relational climate; girls had higher social competence than boys. Although not directly associated with social competence, observed teacher-child interaction quality was conditionally predictive of the social competence in the context of supportive teacher-child relationships. Further, when observed teacher-child interaction quality was average or high, children with low emotional competence exhibited greater growth in social competence than children with high emotional competence. The results inform our understanding of SEC development, the nature of teacher-child interactions in preschool classrooms serving high-risk populations, and potential school-based mechanisms for promoting social competence.