Continuous interscalene brachial plexus block with a single dose of 0.75% bupivacaine (150-210 mg) with adrenaline, continued with an infusion of plain 0.25% bupivacaine 0.25 mg/kg/h, was performed on 20 patients to provide analgesia during shoulder surgery and in the postoperative period. The control group included 20 patients who were given general anaesthesia for surgery after starting a continuous interscalene brachial plexus block; test dose of 0.75% bupivacaine (22.5 mg) with adrenaline, continued with an infusion of 0.25% bupivacaine 0.25 mg/kg/h. Surgery was performed successfully under regional anaesthesia in 16/20 patients; 4/16 were given one dose of fentanyl during the surgery, and diazepam or midazolam as supplementary sedation were given in 13/16 cases. For postoperative analgesia 35/40 patients had a fully functioning catheter for 20-26 hours and the need for oxycodone i.m. during that time was 1.5 +/- 0.4 doses after regional anaesthesia (n = 14) and 1.8 +/- 0.4 doses after general anaesthesia (n = 18). There was a statistically significant difference in the mean plasma bupivacaine concentrations between the groups, concentrations in the regional anaesthesia group being higher at 5, 30, 60 min and 3 h (maximum 2.3 micrograms/ml at 60 min), but there was no difference between the values at 24 h. One infusion of local anaesthetic was discontinued because of probable treatment-related side-effects (breathing difficulties, nausea). Mild local anaesthetic toxicity (dizziness, tinnitus) was noticed in four patients.