Purpose. Conventional approaches to offender profiling assume a homology of the characteristics of offenders with their crime scene actions: the more similar two offenders are with respect to background characteristics, the higher the resemblance in their crime scene behaviour. This implicit working hypothesis is tested empirically.Methods. The study is based on a sample of 100 British male stranger rapists. These individuals were indexed with respect to the similarity in their crime scene actions as derived from witness statements. They were then compared with respect to their socio‐demographic features and criminal histories as derived from police records. In a correlational analysis, we tested whether increased similarity in one domain (offence behaviour) coincided with higher resemblance in the other domains (socio‐demographic features and previous convictions).Results. There is no positive linear relationship for any of the comparisons, i.e. rapists who offend in a similar fashion are not more similar with respect to age, socio‐demographic features (such as employment situation and ethnicity) or their criminal records.Conclusions. These findings indicate no evidence for the assumption of a homology between crime scene actions and background characteristics for the rapists in the sample. We argue that this result suggests that thehomology assumptionis too simplistic to provide a basis for offender profiling. Implications for future research include the search for a suitable framework for offender profiling that is grounded in personality psychology. Further, methodological considerations are discussed, such as the potential application of probabilistic s