Scores from 13 cognitive-neuropsychological tests administered to 245 men alcoholics were subjected to an average linkage cluster analysis. Six subtypes were derived but three were eliminated from further analyses because they had samples of less than 10. The remaining clusters appeared to reflect differing levels and patterns of neuropsychological functioning. Subsequent analyses indicated that the clusters were related to age but were independent of most other demographic and pretreatment alcohol-related variables. Significant associations were found between cluster membership and the likelihood of abstinence, completion of aftercare therapy and employment status over a 9-month follow-up period. Although this relationship to outcome is encouraging, neuropsychological status accounted for only a limited amount of the variance. The findings suggest the need to broaden the base of assessment, using multiple domains, in future attempts to derive meaningful subtypes within an alcoholic population.