A national random sample of 25 percent of the graduate education program directors in internal medicine, family medicine, surgery, and pediatrics was sent a questionnaire; subjects were asked to judge the importance of 31 variables in the selection of house staff. A rank-ordering of variables for all respondents placed interpersonal skills demonstrated in the interview as number one. When rank order correlations were calculated for all possible pairs of program specializations, strong positive relationships were revealed. A two-way analysis of covariance was also undertaken to assess how selected program characteristics, such as size of program, type of program, and affiliation or nonaffiliation with a medical school, affected the judgment of the importance of the variables. The results have implications for further studies in several areas.