This study aimed to analyze the fit of the four-factor model of the 29-item and 12-item Buss-Perry Aggression Questionnaire (BPAQ) on a sample of Portuguese male perpetrators of intimate partner violence (IPV; N = 279) and to test the measurement invariance between sample type (forensic and non-forensic) and sentence type (prison and community). Confirmatory factor analysis (CFA) did not support the four-factor model of the BPAQ 29-item version. The four-factor model of the BPAQ–short form (BPAQ-SF) showed a good fit. Sufferable values for internal consistency and good values for convergent and discriminant validity were found. Cross-sentence measurement invariance using the present sample and cross-sample measurement invariance of the BPAQ-SF using a previously collected sample of students was established. Overall, our results support BPAQ-SF as a valuable brief and psychometrically reliable measure of aggression for use with both individuals from the community and IPV perpetrators with different sentences.