One hundred female patients scheduled for elective breast surgery (mean age 60 +/- 11 years were randomly assigned to receive one of two premedications: ketobemidone (Ketogan) 1-1.5 ml or midazolam 4-5 mg, intramuscularly. The effects on preoperative anxiety and postoperative emetic sequelae were studied. All patients were anaesthetised with thiopentone, fentanyl and atracurium, and ventilated with a mixture of nitrous oxide in oxygen with supplementary isoflurane. Sixty-nine percent of the midazolam- and 50% of the ketobemidone-premedicated patients experienced a reduction in anxiety. Midazolam was found to be more effective than ketobemidone in reducing anxiety among more tense patients--those with a VAS grading before premedication of 2 or more (P less than 0.05). Midazolam-premedicated patients were also assessed by observers as being more relaxed (P less than 0.05). No difference was seen in the frequency of emetic sequelae: 20 patients in the midazolam group and 14 patients in the ketobemidone group vomited once or more during the 24-h observation period. There was no difference between the two groups in time until an analgetic was required. In conclusion, midazolam seemed more effective in reducing preoperative anxiety than ketobemidone without any negative effects on postoperative emesis or time until an analgetic was required.