An attributional intervention was designed to reduce aggressive males' tendency to attribute hostile intentions to peers following ambiguously caused peer provocations. African-American elementary school boys (N = 101), aggressive and nonaggressive, were randomly assigned to the attributional intervention, an attention training program, or a no-treatment control group. Data were collected on subjects' attributions about hypothetical and laboratory simulations of peer provocation, disciplinary referrals to the school office, and teacher ratings of aggressive behavior. Aggressive subjects in the attributional intervention were less likely to presume hostile intent by peers in hypothetical and laboratory simulations of ambiguous provocation. They were also less likely to endorse hostile retaliation on judgment measures and to engage in verbally hostile behaviors in the laboratory task. Further, intervention subjects were rated as less aggressive by their teachers following the treatment. Both the benefits of attributional change and its limitations in the African-American population are discussed.