Objective To assess the knowledge and skill base of the physicians in our hospital in prescribing parenteral nutrition support.Design Physicians completed parenteral nutrition orders for three mock patients. A range of clinically appropriate responses was established before the survey by the hospital Nutrition Support Team. Each case was scored according to set criteria, and the three case scores were averaged to yield a total score of 0 to 100.Subjects/setting Forty-eight attending, resident, and student physicians in the departments of family practice, internal medicine, and general surgery in a community teaching hospital.Statistical analysis Student's t test was used to compare interval data between two groups, and analysis of variance was used to compare interval data among three or more groups. If Bartlett's test for homogeneity of variance showed differences between the groups, the Kruskal-Wallis one-way analysis of variance was used. If analysis of variance revealed a significant difference, the data were reanalyzed using three different multiple-comparison procedures (Tukey, Scheffe, and Bonferroni) to decrease the possibility of a type I error.Results The mean total score was 48.6+/-20.8. Total score differed according to the number of nutrition lectures attended during residency. Total score tended to differ by specialty and number of patients treated with parenteral nutrition; however, these differences did not reach statistical significance.Applications/conclusions This tool can be adapted and used by other hospitals, medical schools, and residency programs to assess physicians' knowledge base, design educational programs, and improve the parenteral nutrition ordering process. J Am Diet Assoc. 1997; 97: 856-859.