The effects of triazolam (0.125, 0.25, and 0.5 mg) versus placebo on recovery sleep staging, subsequent alertness and psychomotor performance were evaluated in humans. Forty-five healthy male subjects were deprived of sleep for 24 h, then administered a single dose of triazolam or placebo using a double-blind procedure. Subjects then attempted to obtain recovery sleep under non-sleep-conducive conditions (sitting upright in a well-lit, crowded chamber) for the next 6 h, followed by 18 more hours of sleep deprivation. During all sleep deprivation periods subjects were tested bihourly on a performance assessment battery which included symbol digit modalities tests (SDMT), four-letter search (FLS), logical reasoning (LR), time estimation (TE), visual vigilance (VV), and short term memory (STM) tasks. Sleepiness levels were measured objectively with multiple sleep latency tests (MSLT) and subjectively with the Stanford Sleepiness Scale (SSS). Compared to placebo, all doses of triazolam resulted in increased amounts of stage 3-4 sleep, and the 0.5 mg dose significantly reduced awakenings (Ps less than 0.05). Although subjects receiving triazolam averaged 21-42 min more total sleep time (TST) than subjects receiving placebo, differences in TST were not statistically significant. Apparent triazolam-mediated benefits to sleep quality resulted in no obvious improvements in performance or alertness levels during subsequent sleep deprivation. It was concluded that the increases in stage 3-4 sleep amounts were most likely due to triazolam-mediated increases arousal thresholds, and the triazolam mediated changes in sleep parameters obtained in the present study were not indicative of substantial changes in the recuperative value of sleep.