We conducted a statewide controlled trial of three methods to improve antibiotic prescribing in office practice: a mailed brochure, a drug educator visit, and a physician visit. Educational topics were three antibiotics contraindicated for office practice and oral cephalosporins. Medicaid prescribing data were used to select donors who needed education. The effect of the methods was evaluated by comparing the change in prescribing (the year before the intervention v the year after the intervention) for the doctors receiving education with the prescribing of comparable doctors chosen as controls. The mailed brochure had no detectable effect, and the drug educator had only a modest effect. The physician visits produced strong attributable reductions in prescribing of both drug classes. For the contraindicated antibiotics, the reductions were 18% in number of doctors prescribing, 44% in number of patients per doctor receiving these drugs, and 54% in number of prescriptions written per doctor. For the oral cephalosporins, both number of patients and number of prescriptions per doctor were reduced by 21%. Doctors responded equally well to recommendations designed to improve the quality of care and to reduce the cost of care.