Nitrous oxide (N2O) effects were evaluated, as a behavioural model of inert gas narcosis, in rats performing a vigilance task in which they were required to respond to a slight luminous increment of the house-light. A dose-related decrease of correct responses was produced by the tested range of concentrations of N2O (from 30 to 70% N2O), with a parallel increase of anticipatory responses up to 60% N2O and a dramatic increase of omissions at 70% N2O. The influence of preparatory period duration on hits delays suggested a 'conditional probability effect' conserved under N2O. Pretreatment of animals with flumazenil (a central benzodiazepine receptor antagonist) did not improve performance under 40% N2O. Naltrexone, an opiate receptor antagonist, however, increased the N2O-induced deficits. These results suggest that benzodiazepine receptors are not involved in N2O-induced impairments in this attentional task, but opiate mechanisms may be implicated.