The current study sought to identify the variables, derived from the self-medication hypothesis, which predicted substance abuse evolution during a homogeneous 3-month antipsychotic treatment. Twenty-four patients were diagnosed with schizophrenia and substance abuse (mainly cannabis and alcohol). Substance abuse, psychiatric symptoms, anhedonia, and social adjustment were assessed at baseline and study endpoint. Linear regression analyses were performed. Better social adaptation and worse anhedonia predicted substance abuse improvements. Conversely, greater psychoactive substance (PAS) use predicted endpoint positive and depressive symptoms. These results suggest that: (i) substance abuse interferes with psychiatric prognosis in schizophrenia; and (ii) dual diagnosis treatments leading patients to engage in alternative social activities may render substance abuse less appealing. Further studies are warranted to dissociate the causes and consequences of substance abuse in schizophrenia.