Understanding classroom-level correlates of preschool children's aggressive behavior is critical to identifying multiple avenues for intervention within schools. The present school-based study evaluated the reliability and validity of a classroom-level measure of physical and relational aggression and examined a social-ecological model to test whether individual variables (i.e., temperament), dyadic peer factors (i.e., peer victimization, the number of a child's play partners), and classroom-level aggression were associated with individual aggression. Observations of play partners and teacher reports of temperament (i.e., daring, prosociality, and negative emotionality), peer victimization (i.e., physical and relational victimization), and classroom aggression were collected in a sample of preschoolers (N = 307; M[subscript age] = 48.99 months, SD = 7.51). Observer reports of aggression were used to create aggression severity and directionality scores, reflecting the overall level of aggression a child displays and their propensity to use physical relative to relational aggression, respectively. There was evidence to support the reliability and validity of the classroom-level aggression measure providing initial support that this measure could be used by school psychologists. A multilevel regression model indicated that higher levels of negative emotionality, daring, and a greater number of play partners were associated with higher levels of aggression severity. Children in classrooms with more relational aggression were more likely to use relational instead of physical aggression. These findings demonstrate the importance for school psychologists to account for multiple levels of influence when examining preschoolers' aggression.