Normal sleepers underwent sleep recordings and daytime tests of sleep tendency, performance, and mood while being shifted 180 degrees in their sleep-wake schedule. After two baseline 24-hour periods, subjects postponed sleep until noon. For the next three 24-hour periods, they were in bed from 1200 to 2000 and received triazolam, flurazepam, or placebo at bedtime in parallel groups. Placebo subjects showed significant sleep loss after the shift. Active medication reversed this sleep loss. Despite good sleep, flurazepam subjects appeared most impaired of the three groups on objective assessments of waking function; triazolam subjects were least impaired.