Investigates production and cost effects of teaching in radiology departments. Models show that cost effects are determined by salaries and fees paid to students and physicians, the levels of student inputs in production, and productivity of student and nonstudent inputs. If students are substitutes for physicians, production costs may be less in teaching than in nonteaching hospitals for a given level of output. Empirical results for Veterans Administration hospitals suggest that teaching reduces costs for most radiology procedures. If teaching can reduce costs of primary products, teaching hospitals may be able to provide a given program of patient care at lower costs than nonteaching hospitals. However, teaching hospitals may provide different treatment patterns than nonteaching hospitals because of differences in case mix, medical techniques, or quality of care. Thus average cost per patient day or per episode of care may still be higher at teaching hospitals than at nonteaching hospitals. 31 pp. Ref.