Homicidal sex offenders represent an understudied population in the forensic literature. Forty-eight homicidal sex offenders assessed between 1982 and 1992 were studied in relation to a comparison group of incest offenders. Historical features, commonly used psychological inventories, criminal histories, phallometric assessments, and DSM diagnoses were collected on each group. The homicidal sex offenders, compared with the incest offenders, self-reported that they had more frequently been removed from their homes during childhood and had more violence and forensic psychiatric contact in their histories. On the self-report psychological inventories, the homicidal sex offenders portrayed themselves as functioning significantly better in the areas of sexuality (Derogatis Sexual Functioning Inventory) and aggression/hostility (Buss-Durkee Hostility Inventory). However, on the Psychopathy Checklist-Revised (PCL-R), researchers rated the homiciders significantly more psychopathic than the incest offenders on Factor 1 (personality traits) and Factor 2 (antisocial history). Police records revealed the homicidal subjects also had been charged or convicted of more violent and nonviolent nonsexual offenses. The phallometric assessments indicated that the homicidal sex offenders demonstrated higher levels of response to pedophilic stimuli and were significantly more aroused to stimuli depicting assaultive acts to children, relative to the incest offenders. Despite the homiciders' self-reports of fairly good psychological functioning, DSM-III diagnoses reliably discriminated between the groups. A large number of homicidal sex offenders were diagnosed as suffering from psychosis, antisocial personality disorder, paraphilias, sexual sadism, sexual sadism with pedophilia, and substance abuse. Seventy-five percent of the homicidal sex offenders had three or more diagnoses compared with six percent of the incest offenders. The article addresses the role of hard versus soft measures in the assessment and treatment of violent sex offenders. In addition, the usefulness of phallometric assessments and the PCL-R and its subscales are considered.