Résumé : Background. This randomized, double-blind study tested the hypothesis that, in comparison with midazolam, premedication with oral clonidine reduces the incidence of emergence agitation in preschool children anaesthetized with sevoflurane. Methods. Sixty-eight ASA I-II children undergoing circumcision were randomized into three groups to receive different oral premedication given 30 min before anaesthesia: midazolam 0.5 mg kg-1, clonidine 2 μg kg-1, and clonidine 4 μg kg-1. Sevoflurane anaesthesia was administered via a facemask (O2/ N2O: 40/60). Analgesia was with penile block (bupivacaine 0.5% 0.3 ml kg-1) and rectal paracetamol (30 mg kg-1). During the first postoperative hour, children were evaluated using a modified 'objective pain scale'. Results. Only the 4 μg kg-1 dose of clonidine was associated with a significant reduction in emergence agitation. Fewer children in the clonidine 4 μg kg-1 group displayed agitation (25%) than in the midazolam group (60%) (P = 0.025). Incidence of hypotension and bradycardia, time to first micturition and first drink did not differ among groups. Conclusions. In comparison with midazolam, clonidine 4 μg kg-1 reduced sevoflurane-induced emergence agitation without increasing postoperative side-effects. © The Board of Management and Trustees of the British Journal of Anaesthesia 2007. All rights reserved.