The Buss-Perry Aggression Questionnaire (AQ) is often used with sexual offenders, but its factor structure has never been examined in this population. The primary aim of this study was to assess the fit of the proposed four-factor model of the AQ reported in previous studies on a sample of incarcerated sexual offenders (N = 293). Results of a series of confirmatory factor analyses did not clearly support the four-factor structure of the full or short version of the AQ. Specifically, very large latent factor correlations suggested that the AQ may not measure a four-dimensional construct in the current sample. Only the physical aggression subscale was independently associated with estimated risk of sexual recidivism. Our findings suggest that the AQ is relevant to risk of sexual recidivism but call into question the appropriateness of the established subscales and their interpretation for sexual offenders.