AIMS: To assess to what extent use of various criteria for high-risk groups and analyses from subpopulations with different drinking patterns may affect the extent of the prevention paradox (that most alcohol-related harm in populations arises within the drinkers at low risk). Data sets Two national surveys of Norwegian adult samples (n = 4321 current drinkers) and one register linkage of Swedish armed forces conscripts (n = 45 839 current drinkers) with in-patient hospital data. MEASURES: High-risk groups were categorized as the upper 10% of drinkers by annual alcohol intake or by intoxication frequency. Acute alcohol-related harms comprised number of quarrels and fights in the Norwegian surveys and number of hospital admissions for attempted suicide and violent injuries over a follow-up period (3 and 25 years) in the Swedish conscript study. RESULTS: The majority of acute alcohol problems were found among the majority of drinkers with low or moderate risk (the lower 90%) by drinking volume, suggesting empirical support for the prevention paradox. By applying frequency of intoxication rather than annual volume of consumption to determine the high-risk group, a somewhat larger proportion of acute alcohol-related harms was found within the high-risk group, and the number of alcohol-related harms tended to be distributed more evenly between high-risk drinkers and other drinkers. The proportion of alcohol-related harms within the risk groups was significantly lower in the younger age group, where the majority drinks to intoxication compared with other drinkers. CONCLUSION: The extent of the prevention paradox with respect to acute alcohol problems may be more prominent in drinking in subpopulations where intoxication is a common part of the drinking pattern compared with those where intoxication occurs less frequently and among a smaller fraction of the drinkers.