Previous studies indicate that haloperidol, a therapeutically useful antipsychotic drug, inhibits neuronal N-methyl-D-aspartate (NMDA) responses and has neuroprotective effects against NMDA-induced brain injury. To further characterize this inhibition, we used electrical recordings to assay the effects of haloperidol on four diheteromeric subunit combinations of cloned rat NMDA receptors expressed in Xenopus laevis oocytes: NR1A coexpressed with NR2A, NR2B, NR2C, or NR2D. Haloperidol selectively blocks NR1A/2B subunit combinations (IC50 = approximately 3 microM; maximum inhibition, approximately 85%), whereas the other subunit combinations are > or = 100-fold less sensitive (IC50 = >300 microM). Inhibition of NR1A/2B receptors is insurmountable with respect to glutamate and glycine and does not exhibit voltage dependence. The splice variant combinations NR1B/2B and NR1e/2B are also blocked by haloperidol. In oocytes from some frogs, 30-100 microM haloperidol induces potentiation of NR1A/2A receptor responses. NMDA responses in E16-17 rat cortical neurons cultured for < or = 10 days are inhibited by haloperidol at the same potency and to the extent as NR1/2B receptors (IC50 = approximately 2 microM; maximum inhibition, approximately 80%). In contrast, cells cultured for longer periods show a wide range of sensitivity. This change in pharmacology coincides with a developmental switch in subunit expression; from NR1 expressed with NR2B to NR1 coexpressed with NR2A and NR2B. Inhibition of macroscopic neuronal NMDA responses is mechanistically similar to inhibition of NR1A/2B receptors. Single-channel recordings from neurons show that antagonism is associated with a decrease in the frequency of channel openings and a shortening of mean channel open time. Collectively, our experiments indicate that haloperidol selectively inhibits NMDA receptors comprised of NR1 and NR2B subunits. Inhibition is consistent with action at a noncompetitive allosteric site that is distinct from the glutamate-, glycine-, and phencyclidine-binding sites and is probably mechanistically related to the atypical antagonist ifenprodil. Our results suggest that haloperidol can be used as a tool for investigating NMDA receptor subunit composition and can serve as a structural lead for designing novel subtype-selective NMDA receptor ligands.