This meta-analysis investigated the possibility that gender differences in aggression, and the variability in these differences, are a function of gender differences in the regulation of arousal generated in emotionally evocative contexts. The sample of studies for this analysis was based on an exhaustive search of the relevant research reports from 1965–1999. Studies were excluded from the sample if they were case studies; investigated spousal/familial or societal violence, war, suicide, or political violence; involved clinical or deviant participants; included fewer than 10 participants; included all male, all female, all non-Caucasian, or non-US/non-Canadian participants. Based on previous evidence that males may be more easily aroused by aggressive-relevant emotional stimuli than females, and that males may have more difficulty regulating emotionally arousing states than females, we hypothesized that the magnitude of the gender differences in aggression would covary, in a nonlinear manner, with the emotional evocativeness of the study context. Consistent with our hypothesis, the magnitude of gender differences in aggression was relatively small in research contexts that appeared to produce no or large increments in emotional arousal and larger (favoring males) in contexts that appeared to produce small or medium increments in emotional arousal. Aggr. Behav. 28:366–393, 2002. © 2002 Wiley-Liss, Inc.