Although alcohol consumption and injury has received a great deal of attention in the literature, less is known about patient's causal attribution of the injury event to their drinking or factors which modify attribution. Hierarchical linear modeling is used to analyze the relationships of the volume of alcohol consumed prior to injury and feeling drunk at the time of the event with causal attribution, as well as the association of aggregate individual-level and socio-cultural variables on these relationships. Data analyzed are from 1955 ER patients who reported drinking prior to injury included in 35 ERs from 24 studies covering 15 countries from the combined Emergency Room Collaborative Alcohol Analysis Project (ERCAAP) and the WHO Collaborative Study on Alcohol and Injuries. Half of those patients drinking prior to injury attributed a causal association of their injury with alcohol consumption, but the rate of causal attribution varied significantly across studies. When controlling for gender and age, the volume of alcohol consumed and feeling drunk (controlling for volume) were both significantly predictive of attribution and this did not vary across studies. Those who drink at least weekly were less likely to attribute causality at a low volume level, but more likely at high volume levels than less frequent drinkers. Attribution of causality was also less likely at low volume levels in those societies with low detrimental drinking patterns, but more likely at high volume levels or when feeling drunk compared to societies with high detrimental drinking patterns. These findings have important implications for brief intervention in the ER if motivation to change drinking behavior is greater among those attributing a causal association of their drinking with injury.