This investigation used two-part growth modeling and cross-lagged panel analysis to examine the predictive function of parenting and teacher-child relationship on the likelihood of children showing problems with parent-rated physical aggression, and on the severity of problems, for 374 children followed from prekindergarten and first grade. Two-part modeling found that teacher-child relationship did not differentiate children who did or did not show aggression problems; parental warmth did, but only in prekindergarten. For children who showed problems with aggression, parental warmth predicted the severity of those problems in prekindergarten, and teacher-child conflict predicted severity of aggression problems in first grade. Cross-lagged panel analyses similarly indicated that parental warmth in prekindergarten predicted aggression in kindergarten, but that kindergarten teacher-child conflict predicted subsequent higher aggression in first grade. Shifts in the importance of specific microsystems over time on children's social development (chronosystem) are discussed, as are the implications for teachers and preservice teacher training.