Previous research on physician utilization has shown that variables found significant in many traditional social psychologic studies (process models) often lack predictive strength if utilized in multivariate research (prediction models). To address this problem, a variable was created that measures both the presence of symptoms and the person’s own evaluation of the necessity for medical care for the symptoms experienced. This variable is a relatively strong predictor of physician utilization. Analysis of findings suggests that future multivariate research on physician utilization consider a focus upon respondent evaluated symptoms (RES) and type of health care facility available. © J. B. Lippincott Co.