Nitrous oxide (N2O, also known as laughing gas) and volatile anesthetics (VAs), the original and still most widely used general anesthetics, produce anesthesia by ill-defined mechanisms. Electrophysiological experiments in vertebrate neurons have suggested that N 2O and VAs may act by distinct mechanisms; N2O antagonizes the N-methyl-D-aspartate (NMDA) subtype of glutamate receptors, whereas VAs alter the function of a variety of other synaptic proteins. However, no genetic or pharmacological experiments have demonstrated that any of these in vitro actions are responsible for the behavioral effects of either class of anesthetics. By using genetic tools in Caenorhabditis elegans, we tested whether the action of N 2 O requires the NMDA receptor in vivo and whether its mechanism is shared by VAs. Distinct from the action of VAs, N 2O produced behavioral defects highly specific and characteristic of that produced by loss-of-function mutations in both NMDA and non-NMDA glutamate receptors. A null mutant of nmr-1, which encodes a C. elegans NMDA receptor, was completely resistant to the behavioral effects of N 2O, whereas a non-NMDA receptor-null mutant was normally sensitive. The N 2O-resistant nmr-1(null) mutant was not resistant to VAs. Likewise, VA-resistant mutants had wild-type sensitivity to N 2O. Thus, the behavioral effects of N 2O require the NMDA receptor NMR-1, consistent with the hypothesis formed from vertebrate electrophysiological data that a major target of N 2O is the NMDA receptor.