Neurocognitive deficits are recognized as core features of schizophrenia and have a great impact on functional outcome. Recent reports have suggested that a functional polymorphism, Val158Met, of the catechol-O-methyltransferase (COMT) gene, partially influences cognitive performances (mainly cognitive flexibility and working memory) both in schizophrenic patients and in healthy controls, probably by modulating prefrontal dopamine function. While previous studies focused on single evaluation of cognitive functioning, we aimed to analyse the additive effect of COMT genotype and cognitive exercise on dynamic modulation of cognitive performances. We analysed the COMT Val158Met polymorphism in 50 patients with chronic schizophrenia randomly allocated to two treatment conditions for 3 months: standard rehabilitation treatment (SRT) alone and SRT plus specific cognitive exercise of impaired functions. We then divided our sample in four subgroups on the basis of genotype (Val/Val versus Met carriers) and treatment (placebo versus active). We assessed patients with a neuropsychological battery, the Positive and Negative Symptoms Scale (PANSS) and the Quality of Life Scale (QLS) at enrolment, after 3 months of therapy and after further 3 months of follow-up. We found significantly greater improvement of cognitive flexibility performance and QLS total score for Met carriers on active treatment in comparison to Val/Val on placebo. The findings support the hypothesis that COMT polymorphism influences individual capacity to recover from cognitive deficit through rehabilitation therapy after a wider intervention also including deficit-specific cognitive exercise as a potentiating tool.