Aims. The methodology of studies that reported data on individuals who recovered from an alcohol or other drug problem (cigarette smokers were excluded) without formal help or treatment were reviewed. Design/ measurements. Potential studies were identified (a) through computerized literature searches, (b) by reviewing references from key publications and (c) by correspondence with researchers in the field. Studies had to (a) be in English, (b) be published, in press, or presented before the end of 1997, (c) report original results or be part of an original survey and (d) separately report respondents whose recoveries were and were not attributable to treatment. No case studies were included. Eligible studies were evaluated with respect to meeting criteria for (a) natural recovery, (b) methodological rigor and (c) reporting demographic and substance abuse history variables. Findings. Until 1997 only 38 articles (40 different respondent samples) met the inclusion criteria for this review. This small number of studies is not surprising, as natural recovery from substance abuse is a relatively new area of study. Moreover, the majority of the 38 articles were published in the past 8 years. For most studies, descriptions of the respondent samples at pre- and post-recovery were seriously deficient. Alcohol was the most studied drug, with heroin a distant second. Low-risk drinking (78.6%) and limited drug use (46.2%) were commonly reported outcomes in natural recovery studies. Conclusions. Based on this review, future natural recovery studies should: (a) report respondents’ demographic characteristics at the time of their recovery; (b) describe respondents’ pre-recovery problem severity; (c) explore in some depth what factors, events or processes are associated with the self-change process; (d) provide corroboration of respondents’ self-reports; (e) examine factors related to the maintenance of recoveries; (f) conduct interviews with individuals who have naturally recovered from cocaine, marijuana and polydrug abuse; (g) include a second interview at a later time to examine stability of natural recoveries; and (h) require a minimum 5-year recovery time frame.