The effect of naloxone, 0.4 mg, on the analgesia induced by nitrous oxide, 33 per cent, in oxygen, was studied in 12 volunteers. Results of previous investigations in animals suggested that endogenous opiate-like substances may play a major role in the analgesic mechanism of nitrous oxide, but the issue had not been studied in man. Cerebral evoked potentials (CEP) to painful tooth-pulp electrical shocks were obtained before and after inhalation of nitrous oxide, and after nitrous oxide plus naloxone, 0.4 mg, in one session; and before and after inhalation of room air, and after room air plus naloxone, 0.4 mg, in another session. CEP waveforms observed between 80 and 350 msec were quantified in terms of three peak-to-peak amplitudes and peak latencies. Nitrous oxide decreased each of the waveform peak-to-peak amplitudes 48 per cent. Naloxone restored the peak-to-peak amplitude of the negative-going wave occurring between 100 and 175 msec. Nitrous oxide also increased the negative peak latency at 175 msec, and naloxone restored this peak latency to normal levels. Neither room air nor room air plus naloxone altered CEP amplitudes or latencies. Over time a significant trend in subjective reports of decreased pain intensity with nitrous oxide and partially increased pain with naloxone was evident. These findings demonstrate that some of the effects of nitrous oxide on the central nervous system can be reversed by naloxone.