This article reports the results of a critical analysis of English-language studies describing programs designed to improve physician prescribing behavior in primary care settings. Only 64% of studies met minimum research design standards, and are the basis for the results reported. The overall findings of the review confirm the conclusions of a previous evaluation of hospital-based studies. (1) The simple dissemination of printed educational materials had no detectable effect on prescribing practice when used alone in well-controlled studies. However, such educational materials do lay a necessary foundation for more personalized educational efforts. (2) Merely distributing computerized listings of patient-specific medication profiles, without specific recommendations for change, is ineffective in reducing overall drug costs or use of inappropriate drugs. Their lack of effect may be due to the large quantity and clinical irrelevance of such data. (3) Continuous computerized reminder systems have been shown to prevent omission of essential care, although such systems are effective only as long as reminders continue. This strategy has not been evaluated for its potential to reduce excessive or inappropriate drug prescribing. (4) Repeated feedback to physicians with specific suggestions for alternative agents has been shown to increase use of generic drugs in academic group practice settings. No adequate studies have tested whether this approach might reduce other types of misprescribing. Also in academic primary care settings, small group tutorials have been shown to improve use of antibiotics and hypertension treatment control. (5) A substantial literature in the U.S. and Canada supports the conclusion that brief, face-to-face educational outreach visits, conducted by either specially trained clinical pharmacists or physician "counsellors" and "opinion leaders", are effective in reducing prescribing of inefficient or contraindicated drugs in pediatric and adult primary care settings. These face-to-face approaches emphasize a tactful, supportive and facilitative role; in some cases, emphasis is put on helping physicians overcome barriers to appropriate prescribing (e.g. patient demand or addiction to minor tranquilizers). Most effective programs also utilize professionally credible sponsors and high-quality educational materials. In addition to improving quality, some programs have been found to produce savings in drug use which exceed the operational costs of the program.