N-methyl-D-aspartate receptor (NMDAR) hypofunction is hypothesised to underlie psychosis but this has not been tested early in illness. To address this, we studied 40 volunteers (21 patients with first-episode psychosis and 19 matched healthy controls) using PET imaging with an NMDAR selective ligand, [18F]GE-179, that binds to the ketamine binding site to index its distribution volume ratio (DVR) and volume of distribution (VT). Hippocampal DVR, but not VT, was significantly lower in patients relative to controls (p = 0.02, Cohen’s d = 0.81; p = 0.15, Cohen’s d = 0.49), and negatively associated with total (rho = −0.47, p = 0.04), depressive (rho = −0.67, p = 0.002), and general symptom severity (rho = −0.74, p < 0.001). Exploratory analyses found no significant differences in other brain regions (anterior cingulate cortex, thalamus, striatum and temporal cortex). These findings are consistent with the NMDAR hypofunction hypothesis and identify the hippocampus as a key locus for relative NMDAR hypofunction, although further studies should test specificity and causality.