One of the main areas of concern regarding offender profiling has been the general lack of extensive empirical studies on the psychological processes underpinning the concept. This study attempts to establish a model of homicide behavior that could be used as a basis for evaluating the scientific validity of offender profiling. A sample of 247 British single offender-single victim solved homicide cases were analyzed using a nonmetric multidimensional scaling procedure known as Smallest Space Analysis. Homicide crime scenes could most readily be differentiated in terms of the expressive and instrumental role of the victim in relation to the offender. Offenders’ backgrounds could similarly be differentiated by an expressive/instrumental thematic split. However, when these two elements were combined, there was a substantial mix between crime scene themes and themes of background characteristics. The article discusses study results in terms of the validity of classifying homicide into an expressive/instrumental dichotomy and possible implications of this classification for offender profiling. Figures, tables, notes, references