Recent epidemiological and social studies have increasingly pointed to the importance of drinking patterns in explaining consequences of alcohol consumption. This paper presents recommendations for research in the area based on the presentations and discussions of the first "International Conference on Social and Health Effects of Different Drinking Patterns" held in Toronto in November 1995. In particular, the social dimension in pattern research, and the relationship between patterns of drinking and casualties as well as social harm, are stressed. The paper also argues for better theories, incorporating knowledge from related basic disciplines. In addition, we emphasize the need for improved methodologies and standardized methods for assessing drinking patterns. Finally, implications of research on drinking patterns for policy and programme development are discussed.