Both the conceptualization and the measurement of satisfaction have been of major concern in medical care research. Some scholars used satisfaction as a determinant of utilization, others as a consequence of utilization. Measures of satisfaction are largely content oriented. Reliabilities are reported, but validity is usually assumed (face validity) or assessed through correlations with a criterion variable like “wish to change providers.” Using multidimensional scaling techniques, this study identifies several components in the construct “satisfaction” and relates them to the utilization of services. A general satisfaction measure provides us with an assessment of physicians and the medical care delivered, a specific satisfaction measure assesses past experience with the regular source of the care. Both of these can be divided into a positive and negative subcluster. Regressions were computed predicting satisfaction from utilization, and utilization from satisfaction, each time controlling in a preceding step for sociodemographic factors. The relationships varied strongly by provider; as a rule, a larger proportion of the variance could be explained analyzing for one provider at a time than for all providers combined. Analyzing one provider at a time showed that satisfaction substantially increased the proportion of explained variance in utilization in hierarchical regression, but utilization did not increase the explained variance in satisfaction.