AIMS: To compare representative general population and treated samples on their reasons for drinking less and whether particular reasons were related to sustained remission from problem drinking for either group. PARTICIPANTS AND DESIGN: A total of 659 problem drinking adults in a Northern California county identified through a probability survey in the general population (n = 239) and a survey of consecutive admissions to public and private alcohol and drug programs (n = 420), who reported drinking less at the one-year follow-up and provided reasons for reducing their drinking, were assessed 1-, 3-, and 5-years post-baseline regarding their problem drinking status. MEASUREMENTS: Logistic regression models were used to predict sustained remission from problem drinking. RESULTS: While the treated sample endorsed a majority of reasons in significantly higher proportions than the general population sample, the same three reasons were significant for both groups in predicting sustained remission from problem drinking: hitting rock bottom, experiencing a traumatic event and undergoing a spiritual awakening. Interventions by medical personnel and family members were either non-significant predictors or significantly negatively related to sustained improvement for both general population and treated problem drinkers. CONCLUSIONS: General population and treatment samples have similar reasons for cutting down as they relate to sustained remission from problem drinking.