BACKGROUND Patient-controlled analgesia (PCA) with intravenous piritramide and subarachnoid bupivacaine was studied during postoperative pain management in comparison with nurse-administered bolus injections. METHODS Following general anaesthesia (n = 60) patients randomly received either 3.75-7.5 mg i.v. piritramide on demand (group P-Bolus) or via PCA (group P-PCA; initial bolus: 3.75 mg i.v. piritramide, baseline rate: 1 mg/h, demand-dose 1.5 mg, lockout time: 20 min). Following continuous spinal anaesthesia (n = 60; CSA; 28-G spinal catheter) patients randomly received a subarachnoid injection of 1.5 ml bupivacaine 0.25% every 2-4 h (group B-Bolus) or a baseline infusion of 0.5 ml/h bupivacaine 0.125% plus 0.5 ml bupivacaine 0.125% on demand via PCA (group B-PCA; lockout time: 30 min). Pain ratings were assessed hourly by patients using a visual analogue scale (0 = no pain, 100 mm = unbearable pain). METHODS multivariate analysis of variance. RESULTS While pain scores did not differ between group P-Bolus and P-PCA, group B-PCA showed the lowest pain ratings (18 +/- 22 mm) differing significantly from group B-Bolus (41 +/- 32 mm; P < 0.001). Group P-PCA required more piritramide than group P-Bolus (46 +/- 15 mg vs. 31 +/- 13 mg, P = 0.001). In contrast group B-PCA required less bupivacaine than group B-Bolus (18 +/- 4 vs. 23 +/- 7 mg, P = 0.002). CONCLUSIONS PCA with CSA was more effective than nurse-administered bolus-administration of bupivacaine, while the present study failed to show superiority of i.v. PCA over i.v. bolus-administration of piritramide. PCA using the subarachnoid route is a promising concept for treatment of postoperative pain in orthopaedic patients, while the PCA piritramide regime of this study warrants improvement.