BACKGROUND Alcohol is a significant risk factor for injuries. This study addresses 1) whether the risk of alcohol related injury increases with frequency of heavy episodic drinking (HED) in a linear fashion, and 2) whether a small group of high risk drinkers accounts for the majority of alcohol related injuries. METHODS We applied a case - control design. Cases were BAC positive injured patients (n = 534) and controls were respondents to a general population survey in Norway (n = 1947). Age and gender adjusted association between self-reported past year HED frequency and alcohol related injury risk was estimated in logistic regression models for all alcohol related injuries and for violence injuries and accident injuries separately. RESULTS An increase in HED was associated with an increase in risk of alcohol related injury, resembling a linear risk function. The small fraction of high risk drinkers (6.6%) accounted for 41.6% of all alcohol related injuries, thus lending support to the validity of the prevention paradox. CONCLUSIONS There is a strong relationship between frequency of heavy episodic drinking and risk of alcohol related injuries, yet the majority of alcohol related injuries are found among drinkers who are not in the high risk group.