Reviews and summarizes findings from a study of numerous surveys to measure patient satisfaction with physicians and medical care services. The authors are critical of many common practices in scoring and interpreting data from such surveys. Specifically: (1) Patient satisfaction is not a unidimensional concept nor can it be adequately treated as a dichotomous variable. (2) Patient satisfaction cannot be reliably nor validly measured with the response to one questionnaire item. (3) Questions that focus on personal satisfaction need to be distinguished from those asking about care received by people in general. (4) The influence of preference and general sentiments on patient satisfaction ratings is not large enough to invalidate satisfaction surveys, and properly constructed surveys will give much information about process and outcomes of care as perceived by consumers. (5) Surveys that demonstrate overwhelming satisfaction with U.S. physicians and medical care are not being interpreted properly.