Varenicline, a partial agonist of the α4β2 nicotinic acetylcholine receptor (α4β2-nAChR), is currently used to facilitate smoking cessation. Preclinical and clinical studies have suggested that this compound may also be effective in treating cognitive impairments in schizophrenia. However, it is unclear which nicotinic acetylcholine receptor subtypes may be involved because varenicline is not only a partial agonist for α4β2-nAChRs but also a full agonist for α7 nicotinic acetylcholine receptors (α7-nAChRs). We investigated the effects of varenicline, compared to the α4β2-nAChR partial agonist TC-2403 and the α7-nAChR full agonist PNU-282987, in a ketamine-based model of schizophrenia-like cognitive deficits on the attentional set-shifting task in rats. The second goal was to elucidate whether the procognitive efficacy of varenicline was due to the compound's action on α4β2-nAChRs or α7-nAChRs. Ketamine was administered to rats for 10 consecutive days and the test was performed 14 days following the last injection. The tested compounds were administered 30 min prior to the attentional set-shifting task. Varenicline, TC-2403 and PNU-282987 ameliorated ketamine-evoked set-shifting deficits. While the α4β2-nAChR antagonist dihydro-β-erythroidine and the α7-nAChR antagonist methyllycaconitine completely prevented the procognitive actions of TC-2403 and PNU-282987, respectively, varenicline's effect was only partially blocked by any given antagonist. Moreover, the combined treatment with TC-2403 and PNU-282987 more effectively facilitated rats' set-shifting ability than activation of either type of nicotinic acetylcholine receptor alone. The present findings demonstrated that varenicline's actions on both α7-nAChRs and α4β2-nAChRs may be necessary to produce its full procognitive effect in the present experimental setting.