Social epidemiological traditions of asking about problems related to drinking are considered. The issue of the attribution of the problem to drinking, and variations in formulations concerning this, are discussed. Social problems from drinking are inherently properties of social interactions, so that they are composed both of behaviour deemed problematic and of a reaction by another. Most items measuring social harm asked of the drinker him/herself are concerned with major social roles, and problems in the particular life area of the role (work, family, friendships, etc.). Some ask the respondent to attribute the problems to alcohol, some ask about others' attributions to alcohol, and some ask about "objective" problem indicators, although these usually have the respondent's attribution to drinking built in. The possibility of a more systematic way of covering different aspects of interactional problems, as reported by the drinker, is considered. Traditions of questioning the person on the other side of the interaction--i.e., items about others' troubles with drinking, and the effect of these on the respondent--are also discussed, and possibilities for bringing questions asked of the drinker and questions asked of interacting others into the same frame are considered.