Birch and Abelson [1] argue that non-income based barriers might explain differences in utilization of health services within and between income groups. Databases which contain utilization data rarely allow for the modelling of geographic variation. In the Ontario Health Survey (OHS), individual observations are georeferenced at the Public Health Unit (PHU) scale, but PHUs cannot easily be used because of the large coefficients of variation. To overcome this problem, a cluster analysis is performed to create a service environment variable, which reflects differences in service availability, population size and rurality. Utilization of health services is then modelled as a logistic regression equation where the independent variables are age, sex, service environment and income to test the Birch and Abelson argument. This argument is then extended by controlling for age, health and income status. Based on the modelling results, the importance of geography to access and utilization is assessed.