An educational intervention designed to change physicians' use of inpatient services was implemented on two general medical services for a year. The intervention consisted of a brief orientation to cost containment issues, a pamphlet that outlined practical cost containment strategies and listed the charges for commonly ordered tests and services, and access to detailed interim patients' bills generated during the hospitalization. Two concurrent control services received no intervention. Over 1,600 admissions were evaluated. The geometric mean length of stay was 0.61 days shorter on intervention services compared with control (5.15 vs. 5.76 days, p less than 0.01). The geometric mean hospital charges were $388 less for intervention patients ($3,199 vs. $3,587, p less than 0.005). Neither patients' demographic characteristics nor case mix could explain the reductions. The authors conclude that a simple program utilizing information already in existence in most hospitals can result in a significant and meaningful reduction in length of stay and charges.