A longitudinal and multilevel approach is used to examine the relationship between antisocial behavior during adolescence and high school social climate. The data are taken from a longitudinal study of 1,233 boys and girls who attended 217 public and private high schools. Students' disruptive behaviors were assessed yearly from 6 to 12 years of age. High school social climate was assessed by teachers, and students reported on their violent and nonviolent antisocial behavior while in high school. The multilevel analyses revealed (1) a large difference between the percentage of variance explained within schools (97%), compared with between schools (3%), and (2) teachers' reports of classroom behavior problems explain between-school differences in student reported antisocial behavior, after controlling for students' family adversity and history of behavior problems during elementary school. The theoretical and practical implications of the study are examined and future directions for research are discussed. (Contains 5 tables and 2 formulas.)