Offender profiling involves the process of predicting the characteristics of an offender based on information available at the crime scene. The hypothesis of this study was that the more similar two offenders were with respect to background characteristics, the higher the resemblance in their crime scene behavior. A sample of 100 British male stranger rapists were indexed with respect to similarity in their crime scene actions, and compared with respect to sociodemographic features and criminal histories. Witness statements and police records were used for data collection. The study tested whether increased similarity in one domain (offense behavior) coincided with higher resemblance in the other domains (sociodemographic features and previous convictions) in a correlational analysis. Results showed that age, sociodemographic features, and previous convictions had no established links with offense behavior. This study has implications for the theoretical underpinning of offender profiling for future research and for practical police work as far as the use of archival data to generate offender profiles is concerned. The conventional approach of assuming a homology of actions and characteristics is likely to fail if there is no clarification of why this should exist. It is possible that some aspects of crime scene behavior are better predictors of characteristics than others. It is also possible that the neglect of situational influences seriously confounds any homology. To control for situational influences as far as possible, future research should separate cases of rape according to situational features, such as whether the offense occurred indoors or outdoors. 2 tables, 6 figures, 34 references, appendix