Within- and between-subject variability in the reinforcing and subjective effects of nitrous oxide (N(2)O) was studied across five sessions. Twelve volunteers with no history of drug dependence sampled 30% N(2)O and 100% oxygen for 10 min each, then chose nine times, once every 5 min, among N(2)O (e.g. "Agent A"), oxygen (e.g. "Agent B"), or "drug-free air." Choice varied across subjects but was stable within subjects. Quantitative differences in subjective effects occurred within and across subjects. Some subjective effects were correlated with choice and/or differed between subjects who were consistent choosers of N(2)O versus those who were not. However, drug liking and euphoria, two face-valid measures of abuse liability, were unrelated to choice. Thus, the present study found individual differences (i.e. between-subject variability) in subjective and reinforcing effects of N(2)O and, in terms of within-subject variability, suggested that subjective effects fluctuate across sessions to a relatively greater extent than do reinforcing effects. The varying degrees of correlation between N(2)O choice and its subjective effects emphasize the need for obtaining multiple measures when characterizing abuse liability of this drug.