There are two conceptual problems with the study of the utilization of ambulatory care: 1) a problem of definition and 2) a problem of measurement. Large multivariate studies conceptualize utilization in a straightforward and static manner that is reflected in the way they measure utilization: as a sum of the number of visits to physicians over a period of time. However, utilization can be construed as a process occurring through time. The object of an analysis of utilization is to examine the causes of the changes in the levels of utilization patients undergo from one period to the next. In this paper, a Markovian model for studying utilization as a process is set up, using data on utilization from a sample of 2,149 patients from Montreal (1981). Results show that the effect of age and sex on utilization is not structured in the same way when using a Markovian model as when using a traditional model, where utilization is measured by the number of visits in one period. Thus, the results of large multivariate studies are most probably biased.