One hundred and ninety-eight children, aged 3 to 12 years, who were scheduled for tonsillectomy were randomly allocated to receive either diclofenac 1.0 mg.kg-1 or papaveretum 0.2 mg.kg-1 by intramuscular injection after induction of anaesthesia. There were no significant differences between the treatment groups in operating theatre blood loss, the frequency of bleeding on the ward, or the need for operative haemostasis. However, the incidence of above average bleeding in the recovery room was significantly higher in the diclofenac group (p < 0.05). Similarly, marked restlessness in the recovery room was more frequent in the diclofenac group (p < 0.01). In both treatment groups there was an association between bleeding and restlessness during recovery so the increased bleeding in the diclofenac group may not be a direct effect.