This study examined the impact of a commonly adopted elementary universal social-emotional learning program, Second Step, on student social-emotional and academic outcomes following 2 years of implementation. The cluster-randomized wait-list control trial included 61 schools, 321 teachers, and 8,941 early elementary students across 2 states, with assessments collected in the fall and spring across 2 years. We developed 8 theoretically derived growth patterns that potentially described the intervention and counterfactual conditions, and then selected the model with the greatest support from the data for each outcome with an information-theoretic approach. Differences between intervention and control groups were tested within the best-fitting model in order to reduce misspecification and maximize generalizability. Most measures fit a pattern in which students improved only during instructional time in school and not during the summer; the data rarely fit the commonly assumed linear growth model. Improvements related to Second Step were observed for emotional symptoms, hyperactivity, skills for learning, and emotion management, but these were modified by either pretest levels or student sex. Condition effects for peer problems were larger for students with initially low ratings and for boys. Boys also showed improvements in conduct problems. Study impacts generally did not depend upon student grade level. The results confirmed the importance of identifying developmentally sensitive trajectories and suggested important implications for Second Step implementation in early elementary grades.