Data were obtained on three kinds of violent offenses reported to the police in Montreal, Canada, during the 1992-1993 period (1,278 sexual assaults, 1,084 noncommercial robberies, and 5,852 other assaults). The analysis was restricted to police files that identified the ethnic status of both offenders and victims and focused on the ethnic composition of violent offenses committed in Montreal's subway stations. The prevalence of interracial violent offenses was higher in subway stations than elsewhere in the urban landscape. Further, the prevalence of interracial offenses was the joint outcome of the differential rate of offending across groups and the structural factors of residential segregation, relative group size, and routine urban mobility patterns. Under certain conditions, urban mobility patterns neutralized the impact of structural factors. Despite obvious similarities between the routine activity approach and the lifestyle exposure model, these analytical frameworks only partially overlapped and were designed for quite different purposes. To the extent that routine activity arrangements alter the impact on crime levels of fundamental social structure parameters, the findings suggest patterns in the circulation of people and property in social space should be added to the limited list of basic or antecedent determinations of aggregate crime distributions. 40 references, 1 note, and 1 table