Background: Converging evidence from neuroimaging and postmortem studies suggests that hippocampal subfields are differentially affected in schizophrenia. Recent studies report dentate gyrus dysfunction in chronic schizophrenia, but the underlying mechanisms remain to be elucidated. Here we sought to examine if this deficit is already present in first-episode psychosis, and if N-methyl-D-aspartate receptor hypofunction, a putative central pathophysiological mechanism in schizophrenia, experimentally induced by ketamine, would result in a similar abnormality. Methods: We applied a mnemonic discrimination task selectively taxing pattern separation in two experiments: 1) a group of 23 first-episode psychosis patients and 23 matched healthy volunteers and 2) a group of 19 healthy volunteers before and during a ketamine challenge (0.27 mg/kg over 10 minutes, then 0.25 mg/kg/hour for 50 minutes, 0.01 mL/s). We calculated response bias–corrected pattern separation and recognition scores. We also examined the relationships between task performance and symptom severity as well as ketamine levels. Results: We report a deficit in pattern separation but not recognition performance in first-episode psychosis patients compared with healthy volunteers (p =.04) and in volunteers during the ketamine challenge compared with baseline (p =.003). Exploratory analyses revealed no correlation between task performance and Repeatable Battery for the Assessment of Neuropsychological Status total scores or positive symptoms in first-episode psychosis patients, or with ketamine serum levels. Conclusions: We observed a mnemonic discrimination deficit but intact recognition in both datasets. Our findings suggest a tentative mechanistic link between dentate gyrus dysfunction in first-episode psychosis and N-methyl-D-aspartate receptor hypofunction.