The hypnotic effect of midazolam on the sleep of pre-surgical patients was studied in 99 patients (53 males, 46 females) due to undergo surgery the following day. One tablet of 15 mg midazolam was administered at 21.00 h and a second was given 45 min later if the subject had not fallen asleep, sleep-onset latency being measured from the time the second tablet was taken. Eighty patients required 1 tablet and 19 required 2 tablets. According to the subjective assessment, patients receiving 1 tablet fell asleep in 22.9 +/- 14.9 min and those taking 2 tablets fell asleep in 38.4 +/- 25.3 min (difference significant P less than 0.05). There was no statistically significant difference between the 1- and 2-tablet groups with regard to sleep duration, number of awakenings, overall assessment of the night's sleep, sleep quality and state on awakening. Factors which had a statistically significant influence on the dosage requirement were (a) sex, 30.4% females requiring a second tablet v. 9.4% males; (b) age, the mean age of the 2-tablet group being 36.5 v. 47 years in the lower dose group; (c) weight, patients with lower body weight requiring the higher dosage, mean 57.5 v. 66 kg; (d) current insomnia or a history of sleeping problems; (e) previous use of hypnotics; (f) degree of insomnia, moderate/severe insomnia needing a higher dosage (42% v. 21%).(ABSTRACT TRUNCATED AT 250 WORDS)