Thirty-six young adult, male subjects with sleep-onset insomnia were equally divided into placebo, 0.25 mg, and 0.5 mg triazolam groups to examine the effects of the hypnotic, with particular attention to dose level on efficacy, sleep stages, and awakening to a smoke detector alarm. On nights 1 and 4 of a five-consecutive-night protocol, a standard home smoke detector alarm was sounded during stage 2, 5 min after sleep onset, in slow wave sleep (SWS), and at the time of the early morning awakening. The alarm registered 78 dB SPL at the pillow. EEG arousal latency and reaction time to a button press were studied. Failure to awaken to three 1-min alarm presentations was scored as "no response." Both dose levels produced similar reductions in sleep latency, decreases in SWS, increases in stage 2, and increases in sleep efficiency. Both dose levels showed similar sedative effects to the smoke alarm. Fifty percent of triazolam subjects failed to awaken on night 1 during SWS, and EEG arousal and response latencies were significantly slowed. Some drug tolerance or sensitization to the alarm was seen by night 4. By morning, all subjects were easily awakened on both nights. The 0.25 mg dose is clearly an effective dose level for both sleep efficacy and sedative effects to outside noise, which in some instances could pose potential problems.