The authors compared the acceptance and efficacy of rectal and nasal administration of midazolam (MDZ) for premedication. Ninety-five ASA I and II paediatric patients (8 months to 12 years) scheduled for elective surgery were randomly allocated to two groups. Group R received 0.3 mg kg-1 of rectal midazolam (in 5 mL saline). Group N received 0.2 mg kg-1 of nasal midazolam (5 mg ml-1). Both groups were divided in two subgroups according to age (group RA (< or = 6 years, n = 33), group RB (> 6 years, n = 18), group NA (< or = 6 years, n = 28), group NB (> 6 years, n = 16)). At the time of premedication, tolerance to the administration was confirmed. Twenty min after rectal or 10 min after nasal administration the quality of sedation was recorded. The nasal midazolam, in commonly used dosages, induced a sedation similar to that following rectal administration with a shorter delay of onset. Nasal administration was more often painful than rectal administration. Swallowing (nasal midazolam) and concerns about modesty (rectal midazolam) were more frequent in older children. Because of its poor tolerance, nasal premedication should be reversed for cases where there is no alternative. Rectal premedication should be avoided in older children.