The present study aimed to assess the pharmaco-clinical profile of infants premedicated with rectal midazolam. The results were compared with those of a reference drug, flunitrazepam. Infants undergoing minor surgery were divided into three groups: group A (n = 30), with a mean age of 15.8 +/- 13.2 months and a mean weight of 8.6 +/- 3.3 kg, receiving 0.33 mg X kg-1 flunitrazepam; group B (n = 15), with a mean age of 11.2 +/- 10.7 months and a mean weight of 9.3 +/- 3.1 kg, receiving 0.3 mg X kg-1 midazolam; and group C (n = 30), with a mean age of 15.5 +/- 9.1 months and a mean weight of 10.7 +/- 2.5 kg, receiving 0.4 mg X kg-1 midazolam. An aqueous solution of each drug was administered with atropine sulfate (0.02 mg X kg-1) 20 min prior to induction of anaesthesia. The drug was well tolerated in 84% of cases. Statistically significant haemodynamic changes consisted of: a 9 c X min-1 decrease in heart rate (p less than 0.05) and a 12 mmHg decrease in systolic and diastolic blood pressures (p less than 0.05) in group C; a 6 mmHg decrease in systolic blood pressure (p less than 0.05) in group B. The tranquilizer action was either excellent or good in 93% of groups B and C compared with only 40% in group A (p less than 0.001 in both cases). Somnolence was attained in 60% of group A, 26.6% of group B and 30% of group C. A mask was much better accepted in group C (86.6%) than in group B (66.6%; p less than 0.05) or in group A (36.6%; p less than 0.01). Therefore, midazolam given rectally at 0.4 mg X kg-1 was better than flunitrazepam, because of a greater therapeutic effect for an equivalent rate of side effects.