This paper reports a case-crossover analysis in a sample of 961 patients who consulted the emergency department (ED) due to an injury in Santa Clara, California, and in Pachuca, Mexico. In the analysis in which usual alcohol consumption during the last 12 months served as the control value, the estimated relative risk of injury in the hour after alcohol consumption, as compared with no alcohol consumption during that time, was 4.33 (CI, 3.55-5.27). After controlling for alcohol use in the 1-h period before injury, the relative risks for consecutive 1-h periods (2-6 h) before the injury were not significantly greater than one, indicating that the induction time was less than 1 h. The relative risk varied greatly depending on race-ethnicity and acculturation among the Hispanics in Santa Clara, with Mexicans in Pachuca showing the highest risk and the high acculturation group in Santa Clara showing the lowest risk. Violence-related injuries were associated with higher relative risk. Relative risk also varied depending on the presence of alcohol dependence and usual frequency of drunkenness: patients with alcohol dependence and patients with high frequency of usual drunkenness had lower risks than patients without alcohol dependence and with lower self-reported episodes of drunkenness in the last year. When blood alcohol content at ED admission was used instead of self-reported alcohol consumption, similar results were obtained. These findings have important public health consequences. Each episode of alcohol consumption results in an increase in the short-term risk for an injury, especially for a violence-related injury. Patients with the lowest usual involvement with alcohol are subject to a higher elevation in their risk for an injury immediately after alcohol consumption compared to patients who drink more heavily.