This Rorschach study explored the suppression of aggression content when violent offenders and nonoffenders are asked to present themselves as not posing a threat of dangerousness in a court role-playing context. Aggressive content and complexity in this suppressive role-play context was compared to a neutral control condition. A total of 41 participants, approximately half violent offenders and half nonoffenders took the Rorschach under both conditions. Results indicate that both groups suppressed aggression content on the Rorschach without altering response complexity. This large effect size for testing condition may partly explain the inconsistencies across previous studies. It is possible that violent offenders have typically been tested in highly suppressive conditions whereas nonoffender or normative groups may have been tested in relatively low suppression conditions. If so, aggression score differences may be a reflection of the testing condition, not group differences. Both instructional sets produced similar levels of complexity, so that individuals do not simplify responses when they screen out aggressive attributions. Violent offenders did not differ from nonviolent offenders in terms of aggression content, but did produce more simplistic records. In addition, this study also undertook a semantic, textual analysis and found that individuals in the suppressive condition tended to eliminate many response elaborations, particularly those with negative of threatening connotations.