International research has suggested that rapist criminal histories can be predicted from certain offence behaviours. Ninety-six solved stranger rape cases were examined to determine if there was consistency in a New Zealand sample. Rapist behaviours during the offence were compared with criminal convictions prior to the rape. The primary statistical technique used was likelihood ratio stepwise binary logistic regression. Predictive utility of the results was limited, due to significant correlation coefficients between the criminal history variables. Nevertheless, the current study reflected the general findings of similar descriptive international research. These outcomes were: the majority of stranger rapists had prior criminal convictions, and the majority of those convictions were for property offences not sexual offending. Furthermore, ethnic minorities were over-represented among the offenders, and the majority of stranger rapists started committing their first known rapes in their midto late 20s. Finally, methodological difficulties were encountered during this study. This highlights the need to refine existing statistical approaches to predictive offender-based research.