OBJECTIVE: To review the results of randomized controlled trials on the effectiveness of brief physician interventions with problem drinkers. DATA SOURCES: The MEDLINE and EMBASE databases were searched for articles published from 1966 and 1972 respectively, with the terms "problem/controlled/responsible/moderate/risk/drink"; "advice/drink"; "physician, nurse, general practitioner"; and "random." Forty-three articles were identified in the EMBASE search and 112 articles in the MEDLINE search. STUDY SELECTION: All trials examining the effectiveness of interventions by physicians in reducing alcohol consumption among problem drinkers attending a health-care facility were reviewed. Trials involving subjects attending an alcohol treatment clinic and those involving interventions delivered solely by nonphysicians were excluded. Eleven trials met the final selection criteria. DATA EXTRACTION: For each article, two of the authors independently assigned a score from 0 to 2 on a number of criteria for validity and generalizability. DATA SYNTHESIS: The four trials with the highest validity scores showed that men in the intervention groups reduced their weekly alcohol consumption by five to seven standard drinks more than the men in the control groups. Results for women were inconsistent. No convincing evidence of declines in alcohol-related morbidity among men or women was found. CONCLUSIONS: The trials support the use of brief interventions by physicians for patients with drinking problems. Although further studies are needed to determine their effect on morbidity and mortality, the public health impact of such interventions is potentially enormous. Further research is needed to determine which patients are best suited for brief interventions, the optimal intensity of treatment and which components of brief interventions are most effective. Research is also needed to establish which strategies are effective in inducing physicians to use brief interventions.