OBJECTIVE: To determine the effect of three types of interaction between physicians and the pharmaceutical industry--company-funded clinical trials, company-sponsored continuing medical education (CME) and information for physicians supplied by pharmaceutical detailers--on orientation and quality of clinical trials, content of CME courses and physicians' prescribing behaviour. DATA SOURCES: MEDLINE and HEALTH searches for English-language articles published from 1978 to 1993, supplemented by material from the author's personal collection. STUDY SELECTION: A total of 227 papers from the MEDLINE and HEALTH searches and about 2000 items from the author's library were initially reviewed. The following selection criteria were used: studies conducted in Australia, Canada, New Zealand, Britain and the United States; studies conducted after 1977; quantitative surveys containing details of the survey methods; studies on the orientation and quality of company-funded clinical trials and on the content of CME courses giving explicit criteria used in the evaluation; and reports on the outcome of interactions stating how the outcomes were assessed. Thirty-six studies met these criteria. DATA EXTRACTION: Information was extracted on five topics: physicians' attitudes toward drug industry interactions, frequency with which physicians participate in the interactions, orientation and quality of company-funded clinical trials, content of company-sponsored CME courses and changes in physicians' prescribing behaviour as a result of an interaction. DATA SYNTHESIS: Although most physicians participate only occasionally in company-sponsored clinical trials, most see detailers and attend company-sponsored CME courses. However, physicians do not have a very high opinion of the information from detailers or of company-sponsored CME events. Many doctors regard pharmaceutical companies as an important source of funding for clinical trials, but they also have concerns about accepting money from this source. Company funding of clinical trials may affect the quality of the trials and the types of research that physicians undertake. Company-sponsored CME courses may have a commercial bias even if conducted under guidelines designed to ensure the independence of the event. All three types of interactions affect physicians' prescribing behaviour and, in the case of obtaining information from detailers, physicians' prescribing practices are less appropriate as a result of the interaction. CONCLUSIONS: Physicians are affected by their interactions with the pharmaceutical industry. Further research needs to be done in most cases to determine whether such interactions lead to more or less appropriate prescribing practices. The CMA's guidelines on this topic should be evaluated to see whether they are effective in controlling physician-industry interactions. Further measures may be necessary if the guidelines fail to prevent negative effects on prescribing practices.