Results indicated that with the exception of violent crime, the ambient-based crime rates had superior goodness of fit. Consistent differences were observed between the ambient-based crime rates and the other crime measures, possibly due to measurement error in the case of residential-based crime rates and a lack of normalization in the case of crime counts. Strong support was found for a routine activities explanation of criminal activity in Vancouver, Canada. The tenets of social disorganization theory, on the other hand, did not hold up well to the statistical model. The analysis relied on a spatial statistical technique to calculate the ambient-based crime rates for automotive theft, breaking and entering, and violent crime, which were then compared to crime rates derived from common measures of crime--crime counts and residential-based crime rates. The ambient population is defined as a 24-hour average estimate of a population in a given area (spatial unit) which can be used to calculate the population at risk. The ambient population for Vancouver, Canada was obtained from the Landscan Global Population Database and the sociodemographic and socioeconomic information was obtained from Statistics Canada’s 1996 Census data. Data on criminal activity in Vancouver was collected from calls for service to the Vancouver Police Department for each month in 1996. Future research using ambient populations should be performed at multiple spatial scales of analysis and should utilize finer crime classifications. Tables, figures, references