In order to test the efficacy of lithium therapy for alcoholism, to test whether lithium therapy reduces depressive symptoms among alcoholics, and to test whether any positive lithium treatment effects are associated with an antidepressant action, we undertook a double-blind placebo-controlled study of 122 patients meeting DSM-III criteria for alcohol dependence and hospitalized for alcoholism rehabilitation. We used survival analysis to examine the cumulative proportion of each treatment group that remained abstinent (i.e., "survived") over the course of 18 months, using complete abstinence from alcohol as the principal outcome measure. Our findings add support to the hypothesis that lithium carbonate promotes abstinence from alcohol and delays the time to first drink. Patients treated with lithium were much less likely to be rehospitalized for alcoholism rehabilitation during the 18-month follow-up. These effects did not appear to be related to the treatment of concomitant or underlying affective symptoms. We found no support, on the other hand, for the hypothesis that lithium therapy reduces the frequency of drinking for relapse drinkers.