OBJECTIVE:Self-reports of alcohol consumption among patients visiting an emergency department (ED) have been used extensively in the investigation of the relationship between drinking and injury. Little is known, however, about the associations between validity of self-reports with patient and injury characteristics and whether these relationships vary across regions or countries. Both of these issues are explored in this article. METHOD:In the construct of a multilevel logistical model, validity of self-reports was estimated as the probability of a positive self-report given a positive blood alcohol concentration (BAC). The setting included 44 EDs across 28 studies in 16 countries. Participants included 10,741 injury patients from the combined Emergency Room Collaborative Alcohol Analysis Project (ERCAAP) and the World Health Organization Collaborative Study of Alcohol and Injuries. Data were analyzed on self-reported drinking within 6 hours before injury compared with BAC results obtained from breath-analyzer readings in all but two studies, which used urine screens. Covariates included demographic, drinking, and injury characteristics and aggregate-level contextual variables. RESULTS:At the individual level, a higher BAC measurement was associated with a higher probability of reporting drinking, as was heavy drinking and sustaining injuries in traffic accidents or violence-related events. At the study level, neither aggregate BAC nor other sociocultural variables affected the validity of self-reported drinking. CONCLUSIONS:This study provides further evidence of the validity of self-reported drinking measures in crossnational ED studies based on the objective criterion of BAC estimates.