Sleep problems and nocturnal arterial oxygen desaturation are common in patients with chronic obstructive pulmonary disease (COPD). Hence, the safety and efficacy of new hypnotic agents must be ascertained in this group of patients. We performed a double-blind, randomized, single-dose, placebo and active drug controlled, crossover study in 24 patients with insomnia (subjective sleep latency > 30 minutes and sleep duration 4-6 hours) and mild to moderate COPD (mean FEV1 61 +/- 12(SD)% predicted) in order to establish the effects of zolpidem 5 mg and 10 mg on sleep and respiration and to compare these effects with triazolam 0.25 mg. Arterial oxygen saturation for the entire night, by hour and stage, and the apnea-hypopnea index for the entire night were not significantly different with placebo and the various drug conditions. Total sleep time and sleep efficiency were increased over placebo by all three drug conditions. Triazolam was more effective than zolpidem 5 mg but not zolpidem 10 mg, and there was no significant difference between zolpidem 5 mg and zolpidem 10 mg. Zolpidem 10 mg and triazolam both reduced the number of awakenings (> 15 seconds duration) per hour of sleep. Although there was a trend for triazolam to be more efficacious than zolpidem 10 mg, no statistically significant difference was found for any objective or subjective sleep variable. Likewise, zolpidem 10 mg tended to be more efficacious than zolpidem 5 mg, but the difference was only significant in terms of perceived sleep quality and ease of falling asleep.(ABSTRACT TRUNCATED AT 250 WORDS)