Sixty-three children under the age of 9 years were randomized to receive intravenous (group A, n = 33) or intranasal (group B, n = 30) midazolam as sedation for small bowel biopsy. Mean doses of midazolam given to produce adequate sedation were 0.31 mg (kg body weight)-1 in group A and 0.34 mg (kg body weight)-1 in group B (NS). Four children in group A and 10 children in group B required additional doses to maintain adequate sedation throughout the biopsy procedure (p < 0.05). There was no significant difference between the groups regarding the median procedure time (7 min in group A, 8.5 min in group B) or median fluoroscopy time (5 s in group A, 4 s in group B). All children in group B showed signs of discomfort from the nose when given midazolam intranasally. In conclusion, this study indicates that intravenous administration of midazolam is preferable to the intranasal route.