In order to conduct a comparative risk analysis for alcohol within the Global Burden of Disease Study (GBD 2000), several questions had to be answered. (1) What are the appropriate dimensions for alcohol consumption and how can they be categorized? The average volume of alcohol and patterns of drinking were selected as dimensions. Both dimensions could be looked upon as continuous but were categorized for practical purposes. The average volume of drinking was categorized into the following categories: abstention; drinking 1 (> 0-19.99 g pure alcohol daily for females, > 0-39.99 g for males); drinking 2 (20-39.99 g for females, 40-59.99 g for males), and drinking 3 (> or =40 g for females, > or =60 g for males). Patterns of drinking were categorized into four levels of detrimental impact based on an optimal scaling analysis of key informant ratings. (2) What is the theoretical minimum for both dimensions? A pattern of regular light drinking (at most 1 drink every day) was selected as theoretical minimum for established market economies for all people above age 45. For all other regions and age groups, the theoretical minimum was set to zero. Potential problems and uncertainties with this selection are discussed. (3) What are the health outcomes for alcohol and how do they relate to the dimensions? Overall, more than 60 disease conditions were identified as being related to alcohol consumption. Most chronic conditions seem to be related to volume only (exceptions are coronary heart disease and ischemic stroke), and most acute conditions seem to be related to volume and patterns. In addition, using methodology based on aggregate data, patterns were relevant for attributing harms for men but not women.