Physicians endorse prevention but provide only low levels of screening, health counseling, and immunization. Between 1981 and 1986, a randomized controlled trial was conducted at the Seattle (Wash) Veterans Affairs Medical Center to assess the effectiveness of the following three methods of delivery of preventive services: (1) a physician-oriented model that includes education and motivation, a chart flowsheet listing recommended activities, and periodic feedback about performance; (2) a patient education model in which patients were mailed an informative brochure advising them to ask physicians for preventive services as depicted in a patient-held pocket guide; and (3) a health promotion clinic that patients were invited to attend. A control group received their usual care. A total of 1224 male outpatients were enrolled in the trial. Baseline prevention rates for 12 age-specific prevention activities were below 25%. Neither the control group rates during the 5-year trial nor the rates for the two educational models, either singly or as a combined intervention, changed. Only the health promotion clinic model was effective, tripling prevention rates in its first year and sustaining these levels for all 5 years. It is difficult to change the clinic roles of experienced physicians and their long-term patients in a specialized multiclinic setting. Providing a separate health promotion clinic option is popular with patients, bypasses gatekeeper barriers, is reasonable in cost, and warrants wider application.