Marlatt and Gordon's (1985) relapse prevention therapy has received widespread interest and application. The categorization of relapse precipitants was one of the original central features of this model. In more recent iterations of this therapy, increasing emphasis has been placed on coping strategies. In the present article, exploratory findings from a prospective naturalistic alcohol treatment study employing the Reasons for Drinking Questionnaire are reported. A relapse precipitants scoring algorithm is presented allowing relapses to be categorized as either negative affect relapses, social pressure relapses, or craving/cued relapses. Exploratory findings suggest that social pressure relapses are more likely to repeat, and that negative affect and craving/cued relapses are more severe. Perhaps most interestingly, craving/cued relapses appear to subside during the first 6 months following treatment initiation, but subsequent risk for this type of relapse returns if the client has relapsed. However, these findings are still early in a continuing exploration of these issues in relapse prevention.