Background: Therapeutic communities (TCs) are drug-free residential settings, focused on psychosocial rehabilitation. While TCs are considered an effective method, the bulk of the research evidence is from poorly controlled studies. The goal of this study was to evaluate TC effectiveness in terms of abstinence and to determine if there were predictive factors of abstinence. Methods: The search used Medline up to January 30, 2011 and was based on a systematic review method. Studies on retention in treatment and/or substance use were considered. Results: Of the 321 studies retrieved from Medline, 12 met selection criteria including overall 3,271 participants from 61 TCs. On average, subjects stayed in TC a third of the planned time. The completion rate ranged from 9 to 56%. All studies showed that substance use decreased during TC, but relapse was frequent after TC. Treatment completion was the most predictive factor of abstinence at follow-up. Surprisingly, psychiatric comorbidities did not appear associated with relapse or with dropout. Conclusions: There was a drop in consumption after TC, but long-lasting benefits were uncertain. Further studies are needed in order to compare the efficacy of TC programs and other types of treatment settings for substance-related disorders.