Novel object recognition (NOR) in rodents is analogous in some ways to human declarative (episodic) memory, one of the seven cognitive domains which are abnormal in schizophrenia. Cognitive impairment in schizophrenia (CIS) accounts for the largest proportion of the poor functional outcomes in this complex syndrome, with psychosis and negative symptoms accounting for much of the rest. Current atypical antipsychotic drugs (APDs) e.g. amisulpride, aripiprazole, clozapine, lurasidone, olanzapine and risperidone, and typical APDs as well, significantly improve some, but not all aspects of CIS, including declarative memory, but not in all patients, and rarely restore normal function. Thus, finding new ways to prevent or treat CIS is a major goal of current schizophrenia research, with animal models as an essential tool. NOR in rodents is valuable in this regard because of its relationship to declarative memory, the extensive knowledge of its underlying circuitry, and the ease and reliability of assessment. Sub-chronic administration of an N-methyl-Daspartate receptor (NMDAR) non-competitive antagonist, e.g. phencyclidine (PCP), dizocilpine (MK-801) or ketamine, is a favored means to study NOR as a model of CIS, because it produces deficient glutamatergic and GABAergic function, both of which have been implicated in the development of CIS. Transgenic mice and anti-cholinergic-induced deficits in NOR have received less attention. We review here NOR studies in rodents that bear upon CIS, including the evidence that atypical, but not typical APDs, as well as specific ligands, e.g. 5-HT1A partial agonists, 5-HT7 antagonists, D1 agonists, among others, can restore NOR following sub-chronic NMDAR antagonist treatment, and can also prevent the impairment in NOR produced by sub-chronic NMDAR antagonists. We discuss how well these findings translate to the bedside.