This study addresses patient attrition in order to identify variables associated with retention in conjoint treatment for alcohol abuse and dependence. Subjects were 105 male alcoholics and their partners who participated in a randomized clinical trial of three approaches to conjoint treatment of alcoholism: (a) alcohol and behavioral marital therapy (ABMT); (b) ABMT plus Alcoholics Anonymous and Alanon (AA/ABMT); and (c) ABMT plus relapse prevention (RP/ABMT). Completers (n = 45) attended two assessment and 15-17 treatment sessions. Partial completers (n = 23) left treatment between the 5th and 15th treatment session. Thirty-seven dropouts left treatment between the first assessment session and the 5th treatment session; of these, 22 dropouts completed at least one treatment session (called early treatment dropouts). Of subjects who entered treatment, dropouts from early treatment were least compliant with treatment requirements in the first few sessions, less educated, and less committed to their relationship. Data indicate that therapists with more clinical experience were more likely to retain clients in treatment, regardless of treatment condition. The data do not reflect a linear relationship between most variables studied and a continuum of attrition status from assessment to completion of the program.