Six healthy human volunteers (ages 18 to 24) acquired a triazolam (0.32mg/70kg) vs placebo discrimination under a standard, two-response drug discrimination procedure. Dose-effect curves were then determined for triazolam (0.1-0.56mg/70kg), lorazepam (0.75-3.0mg/70kg) and buspirone (7.5-30mg/70kg) under a novel response procedure that provided a response alternative for drugs unlike triazolam or placebo (i.e. a novel-appropriate response). Triazolam dose-dependently increased triazolam-appropriate responding but did not occasion any novel-appropriate responding. Lorazepam dose-dependently increased triazolam-appropriate responding in four of six subjects, but at least one dose also occasioned novel-appropriate responding in three subjects. Buspirone dose-dependently increased novel-appropriate responding, although three of six subjects also made triazolam-appropriate responses following some dose(s). All three drugs comparably increased self-reported sedation. Self-reported effects did not differentiate triazolam from lorazepam whereas only buspirone increased "bad" self-reports, and did not increase "liking" and "good" self-reports. The results suggest that the novel response procedure enhanced the pharmacological selectivity of human benzodiazepine discrimination and may help interpret partial generalization under two-choice drug discrimination procedures. The results also add to the evidence of a close relationship between the discriminative stimulus and self-reported effects of drugs.