Sexual sadism is assumed to be a crucial factor in sexual homicide. Prevalence estimates vary greatly due to differences in the definition of sexual sadism. A nationwide sample of 350 male perpetrators who had committed a sexual homicide offense against a female 14 years of age or above in England or Wales was assessed based on archival records. Sexual sadism was assessed using the Sexual Sadism Scale (SeSaS). Item response theory (IRT) analyses were conducted focusing on the 2-parameter logistic model. The single-factor structure of the SeSaS Part 1 was tested using confirmatory factor analysis. Estimates of both internal consistency and interrater agreement were satisfactory to substantial. IRT analysis showed that the Part 1 items captured moderate to severe levels of the latent construct (i.e., theta levels >0). Based on the Posterior Probability of Diagnosis index, the prevalence of the disorder was estimated at 37% in the sample. The substantial correlation between the SeSaS Part 1 total score and original clinical diagnoses of sadism confirms the criterion validity of the scale. Exertion of control and infliction of torture were among the more informative items. In sum, the results support the usefulness of the SeSaS instrument for assessing forensically relevant forms of sadism. (PsycINFO Database Record (c) 2018 APA, all rights reserved).