BACKGROUND: The purpose of this study was to compare the postoperative analgesic effects and side effects of an intra-articular PCA infusion of bupivacaine and morphine using an intravenous PCA infusion of morphine following arthroscopic shoulder surgery. METHODS: Seventy-one patients, undergoing arthroscopic shoulder surgery under general anesthesia, were randomly assigned to one of two groups. In group 1 (n = 32), morphine and ondansetron, 8 and 4 mg, respectively, were intravenously injected following surgery, with the subsequent infusion of normal saline 100 ml, including morphine and ondansetron, 32 and 12 mg, respectively, through an intra-venous PCA catheter. In group 2 (n = 39), 0.25% bupivacaine, 40 ml, including an intra-articular injection of morphine, 3 mg, followed by an infusion of 0.25% bupivacaine, 100 ml, including morphine, 5 mg, were administered through an intra-articular PCA catheter. In groups 1 and 2, the PCA infusion rate was 2 ml/h, with a bolus dose of 0.5 ml, with a lock out time of 8 min. The VAS for pain at rest, and the range of motion (ROM) exercise and side effects were assessed 0.5, 1, 2, 4, 12, 18 and 24 h postoperatively. RESULTS: The patients in group 2 had significantly lower VAS for pain for the ROM than those in group 1 30 min postoperatively. However, the VAS for pain at rest was significantly lower in group 1 than 2 after 18 and 24 h, but the VAS for pain for the ROM was significantly lower in group 1 than 2 24 h postoperatively. There was no significant difference in the side effects between the two groups, with the exception of dizziness, which was more severe in group 2 at 1, 2 and 4 h postoperatively. CONCLUSIONS: An intra-articular PCA infusion of bupivacaine and morphine is no more effective than an intra-venous PCA infusion of morphine and ondansetron with respect to postoperative analgesia and side effects.