The objective of this study is to assess whether the antidepressants are effective in treating major depressive disorder (MDD) in patients with a co-morbid axis-III disorder, and to compare the relative efficacy (vs. placebo) of antidepressants between these patients and those enrolled in general MDD trials. Medline/Pubmed publication databases were searched. We selected for randomized, double-blind, placebo-controlled trials of antidepressants for MDD, whether conducted in a general population, or in populations with an axis-III disorder. Antidepressants were more effective than placebo in patients with axis-III disorders overall [risk ratio for response (RR)=1.42, P<0.0001], as well as specifically for post-stroke (RR=1.43, P=0.04), human-immunodeficiency virus positive or acquired immunodeficiency syndrome (RR=1.66, P=0.005), but not cancer patients (RR=1.26, P=0.19). There was a trend for higher placebo response rates in studies, which did (41.2%) versus those which did not (37.7%) select for axis-III disorders (P=0.06), and significantly higher antidepressant response rates in studies which did (57.4%) versus those which did not (53.5%) select for axis-III disorders (P=0.01). Antidepressants are effective for MDD in patients who present with co-morbid axis-III disorders and as effective (vs. placebo) in this population as they are in the general MDD population. Higher general response rates observed in the co-morbid MDD population are intriguing, and require replication.