To evaluate the effect of a patient, health care provider, and systems intervention on the prevalence of risk factors for lower extremity amputation in patients with non-insulin-dependent diabetes. Blinded, randomized, controlled trial. Academic general medicine practice. Of the 395 patients with non-insulin-dependent diabetes who underwent the initial patient assessment, 352 completed the study. The 12-month intervention was multifaceted. Patients received foot-care education and entered into a behavioral contract for desired self-foot care, which was reinforced through telephone and postcard reminders. Health care providers were given practice guidelines and informational flow sheets on foot-related risk factors for amputation in diabetic patients. In addition, the folders for intervention patients had special identifiers that prompted health care providers to: 1) ask that patients remove their footwear, 2) perform foot examinations, and 3) provide foot-care education. Patients receiving the intervention were less likely than control patients to have serious foot lesions (baseline prevalence, 2.9%; odds ratio, 0.41 [95% CI, 0.16 to 1.00]; P = 0.05) and other dermatologic abnormalities. Also, they were more likely to report appropriate self-foot-care behaviors, to have foot examinations during office visits (68% compared with 28%; P < 0.001), and to receive foot-care education from health care providers (42% compared with 18%; P < 0.001). Physicians assigned to intervention patients were more likely than physicians assigned to control patients to examine patients' feet for ulcers, pulses, and abnormal dermatologic conditions and to refer patients to the podiatry clinic (10.6% compared with 5.0%; P = 0.04). An intervention designed to reduce risk factors for lower extremity amputations positively affected patient self-foot-care behavior as well as the foot care given by health care providers and reduced the prevalence of lower extremity clinical disease in patients with diabetes.