Fifteen subjects (9 men and 6 women) exhibiting objective evidence of excessive daytime somnolence and periodic leg movements in sleep underwent 4-7 days of treatment with triazolam (0.25 or 0.50 mg) and placebo in a double-blind crossover design. One night of polysomnography followed by daytime multiple sleep latency testing were conducted on the first and last days of each treatment block. By the last day of treatment, the mean multiple sleep latency test score after triazolam (9.0 minutes) was significantly greater than that after placebo (5.7 minutes). Thus, triazolam treatment led to a decrease in daytime somnolence. Triazolam also improved sleep architecture and continuity; it increased total sleep time, decreased the number of awakenings and arousals, and decreased stage 1 and increased stage 2 percentages. Although the frequency of periodic electromyographic bursts remained unchanged, the frequency of associated arousals decreased after treatment. Short-term treatment with triazolam is thus effective in diminishing daytime sleepiness and in improving sleep architecture, continuity and duration in patients with periodic leg movements in sleep. These effects do not seem to be mediated through a decrease in periodic leg movement activity.