The concepts of hospital service population and its estimation techniques is refined and generalized from a model-building point of view, and the generalized model is applied to the Alberta, Canada, hospital system. The assumptions underlying the so-called relevance and commitment index methods are investigated. A set of computer programs is developed for estimating the service population of Alberta hospitals. The programs use census, patient origin, and hospital statistics related to costs as input, and provide resource allocation and utilization rates on an age-sex adjusted per capita basis for all Alberta general acute hospitals and hospital districts. The estimates based on relevance and commitment index methods are compared and found to be very similar, except at the extreme tail areas of the distribution.