We examined the effectiveness of a training program for physician-delivered nutrition counseling, alone and in combination with a structured office practice environment for nutrition management, on physicians' counseling practices. Forty-five primary care internists and 1,278 of their patients in the top quarter of the cholesterol distribution at a central Massachusetts health maintenance organization (the Fallon Clinic) were enrolled into a randomized controlled trial. Physicians were randomized by site into three conditions: (1) usual care, (2) physician nutrition counseling training, and (3) physician nutrition counseling training plus a structured office practice environment for nutrition management (prompts and the provision of lipid results and counseling algorithms). A randomly selected 325 patients were given a 10-item patient exit interview (PEI) assessing whether the physician provided advice; assessed past changes, barriers, and resources; negotiated specific plans and goals; provided patient materials; referred the patient to a dietitian; and developed plans for follow-up. Condition 3 physicians demonstrated significantly greater implementation of the nutrition counseling sequence than did physicians in either of the other two conditions (P < .0001). Referrals to nutrition services were markedly reduced in condition 2, despite PEI scores no different than those in condition 1. Higher PEI scores for patients seen by physicians in condition 3 were stable for as long as two years beyond training. Primary care internists, when provided with both training in counseling techniques and a supportive office environment, will carry out patient counseling appropriately. Training alone, however, is not sufficient and may be counterproductive. Medical Subject Headings (MeSH): hypercholesterolemia, diet therapy, coronary disease, health behavior, primary health care, medical education, managed care programs.