Background: People with schizophrenia frequently have cognitive dysfunction, which does not respond to pharmacological interventions. Varenicline has been identified as a potential treatment option for nicotinic receptor dysfunction with a potential to treat cognitive impairment in schizophrenia. Methods: We conducted a systematic review of Pubmed, Embase, Psycinfo, CINAHL and the Cochrane Schizophrenia Trial Registry for randomised controlled trials of varenicline in people with schizophrenia for cognitive dysfunction. We excluded trials among people with dementia. We then undertook a meta-analysis with the primary outcome of difference in change of cognitive measures between varenicline and placebo as well as secondary outcomes of difference in rates of adverse events. We conducted a sensitivity analysis on smoking status and study duration. Results: We included four papers in the meta-analysis (n = 339). Varenicline was not superior to placebo for overall cognition (SMD = −0.022, 95% CI −0.154–0.110; Z = −0.333; p = 0.739), attention (SMD = −0.047, 95% CI −0.199–0.104; Z = −0.613; p = 0.540), executive function (SMD = −0.060, 95% CI −0.469–0.348; Z =− 0.290; p = 0.772) or processing speed (SMD = 0.038, 95% CI −0.232–0.308; Z = 0.279; p = 0.780). There was no difference in psychotic symptoms, but varenicline was associated with higher rates of nausea. Sensitivity analyses for smoking status and study duration did not alter the results. Conclusion: Within the present literature, varenicline does not appear to be a useful target compound for improving cognitive impairment in schizophrenia. Based on these results, a trial would need over 2500 participants to be powered to show statistically significant findings.