This article reports the results of an emergency room study carried out in 1987 in three hospitals in Acapulco, Mexico. All patients were interviewed and breath tested for alcohol consumption. The data were analyzed using the case-control methodology. Cases were falls, motor vehicle accidents, fights and assaults, and home accidents. The control group included patients for whom accidents are less frequently reported as alcohol related, i.e., those reporting work place accidents, animal bites and recreational accidents. For positive breathalyzer readings (> or = 10 mg/100 ml of blood alcohol), the odds ration and 95% confidence intervals were statistically significant for falls, 3.45 (1.23-9.66); motor vehicle accidents, 3.85 (1.21-12.01); and fights and assaults, 5.23 (2.36-12.95). The association was non-significant for home accidents, 0.78 (0.20-2.98). Possible biases of this approach are discussed and recommendations are made for selecting better controls for future emergency room studies.