What are often called "little-ticket" items--X-rays and laboratory tests--account for 25-30% of all health care costs. Two such items were the focus of this study, which took place at an inner-city community health center operated by the Department of Family Medicine of Cook County Hospital and involved 20 family practice residents over a period of nine weeks. The first item was the complete blood count (CBC) with differential, a prototype of a low-cost, high-volume test often ordered by reflex; the second, the thyroid stimulating hormone (TSH) test, a high-cost, low-volume test normally associated with a differential diagnosis or clinical reasoning. Through the use of a simple educational intervention based on quality of care, not cost-containment, and an audit feedback system, the authors were able to reduce significantly the rates of ordering TSH tests (p less than .0001) and CBCs (p = .05). This effect on the rates persisted five months after the intervention terminated. In addition, the percentage of TSH tests clinically indicated by explicit criteria increased significantly (p less than .001) during the intervention. However, this effect showed signs of diminishing five months after the intervention ended. The percentage of CBCs clinically indicated did not change significantly as a result of this intervention.