Empirical studies assessing alcohol’s harm to others very often rely on population survey data. This study addresses some of the problems and challenges in using survey data for this purpose. Such problems include the limited capacity of population surveys in identifying infrequent harm and long-term consequences of drinking. Moreover, the drinker may report the alcohol-related harm or the person being harmed may report the damage. However, irrespective of who reports the harm, causal attribution to drinking is problematic. Challenges for future population surveys to address alcohol’s harm to others include the need for improved models and understanding of complex mechanisms to guide empirical studies within the broad range of harm. Study designs other than cross-sectional surveys, such as longitudinal study designs and combinations of population surveys and other data sources, are likely to overcome some of the identified problems in current population surveys of alcohol’s harm to others.