Previous reviews have concluded that there was no evidence for the superiority of inpatient over outpatient treatment of alcohol abuse, although particular types of patients might be more effectively treated in inpatient settings. In this review, we first consider the conceptual rationales that have been offered to support inpatient and outpatient treatment. Following that, the results of the relevant research on setting effects are presented. Five studies had significant setting effects favoring inpatient treatment, two studies found day hospital to be significantly more effective than inpatient treatment, and seven studies yielded no significant differences on drinking-related outcome variables. In all but one instance in which a significant effect emerged, patients in the 'superior' setting received more intensive treatment and patients were not 'preselected' for their willingness to accept random assignment to treatment in either setting. Studies finding significant setting effects also conducted more treatment contrasts (18.6 vs. 4.9), on average, and had a mean statistical power level of 0.71 (median 0.79) to detect a medium-sized effect, whereas studies with no significant findings had an average power level of 0.55 (median 0.57). When inpatient treatment was found to be more effective, outpatients did not receive a respite in the form of inpatient detoxification and the studies were slightly less likely to have social stability inclusion criteria and to use random assignment to treatment settings. We consider the implications of our findings for future research, especially the need to examine the conceptual rationales put forward by proponents of inpatient and outpatient treatment, i.e. mediators and moderators of setting effects.