Objective: To summarize the scientific evidence about the efficacy of therapeutic communities (TC) to reduce substance use and related problems among people with substance use disorders. Methods: This systematic review builds from the work performed by Smith et al. (2006). We searched MEDLINE, EMBASE, Web of Science, Scielo, and LILACS for randomized trials that compare a TC with no treatment, a different type of treatment or another type of TC published from March 2004 to May 2011. Results: 5 publications from 4 randomized trials were identified. All the studies had serious methodological limitations according to the CONSORT. The heterogeneity among studies did not allow for meta-analytic analysis to calculate pooled estimates. The primary analysis showed that, in prison, certain models of TC might be marginally superior to other types of treatments regarding levels of alcohol use, days in prison and re-incarceration rates. Also, evidence from a community setting (i.e., not in-prison) suggests that a community-based TC is not superior to an outpatient treatment model regarding levels of substance use, crime and unemployment at the 12-month follow-up. Conclusions: In general, there is no evidence to support superiority of TC over other more accessible and less costly types of treatment for drug use. However, in a prison context, TC might be of more benefit than other types of treatment. More research with solid experimental methodology is needed to add to the still weak body of evidence that supports the use of TC over other more affordable types of treatment for drug use disorders.