Although a substance use disorder (SUD) is traditionally associated with psycho-bio-social impairments, recent investigations among persons with schizophrenia (Sz) generated divergent results. Certain persons with Sz+SUD might in fact present better social and cognitive functioning than persons with Sz without SUD. This meta-analysis was conducted to verify this counterintuitive possibility and to determine whether factors such as substance type, severity or nature of psychotic symptoms and age of the patients help discriminate these subgroups. Twenty-three studies met the inclusion criteria and data from 1807 persons with schizophrenia, with or without comorbid SUD, were available for analyses. As a group, persons with Sz+SUD did not obtain significantly higher scores at a Global Cognitive Index than persons with Sz without SUD, although they were better at the Trail Making Task and the speed processing domain. Secondary analyses showed the importance of considering intermediate factors, particularly the preferred substance used and the mean age. While consumption of alcohol was associated with a global cognitive scores similar to that of persons with Sz without an SUD and lower working memory capacities, preferential use of cannabis was instead associated with higher scores for problem solving and reasoning and visual memory. Age was inversely related to the size of the effects. It is concluded that previous mixed results obtained with cognitive evaluations of persons with Sz+SUD might reflect the heterogeneity of participants and that subgroups of patients might be defined on the basis of intermediate factors.