Background Previous studies have shown that patients with schizophrenia have higher smoking rates and worse cognitive function than healthy controls. However, there is no consistent conclusion about the relationship between smoking and cognitive impairment. Objectives The main purpose of this study was to explore the effects of smoking on cognitive function by using MATRICS Cognitive Consensus Battery (MCCB) in Chinese male patients with schizophrenia. Methods There were 164 patients with chronic schizophrenia and 82 healthy controls. All subjects were interviewed about smoking status. The cognitive function was assessed by MCCB and Stroop tests. The Positive and Negative Syndrome Scale (PANSS) was used to assess the clinical symptoms of the patients. Results Compared with healthy controls, patients had lower MCCB scores in all of its domain scores (allp< 0.05). In the patients, the scores of spatial span test (42.3 +/- 11.6), digital sequence test (42.9 +/- 10.6), and Hopkins Verbal Learning Test (42.2 +/- 10.1) were lower in smokers than those in nonsmokers (allp< 0.05, effect size: 0.28-0.45). Logistic regression analysis showed that the smoking status of the patients was correlated with digital sequence score (p< 0.05, OR = 1.072, 95%CI: 1.013-1.134). Multivariate regression analysis showed that the spatial span total score (beta= - 0.26,t= - 2.74,p< 0.001) was associated with the duration of smoking in patients with schizophrenia. Conclusions Our findings show that smoking patients with chronic schizophrenia exhibit more severe cognitive impairment than nonsmoking patients, especially in working memory and executive function.