Background Cyclooxygenase (COX)-2 antagonist is widely used for intravenous postoperative pain relief. Recent studies reported COX-2 in the spinal dorsal horn could modulate spinal nociceptive processes. Epidural parecoxib in rats showed no neurotoxicity. These findings suggested applying a COX-2 antagonist directly to the central or peripheral nerve might provide better analgesia. Questions/purposes We therefore determined: (1) whether the addition of parecoxib to ropivacaine injected locally on the nerve block affected the sensory and motor block times of the brachial plexus nerve block; and (2) whether parecoxib injected locally on the nerve or intravenously had a similar analgesic adjuvant effect.Methods We conducted a randomized controlled trial from January 2009 to November 2010 with 150 patients scheduled for elective forearm surgery, using a multiplenerve stimulation technique. Patients were randomly allocated into one of three groups: Group A (n = 50) received ropivacaine 0.25% alone on the brachial plexus nerve; Group B (n = 50) received ropivacaine together with 20 mg parecoxib locally on the nerve block; and Group C (n = 50) received 20 mg parecoxib intravenously. We recorded the duration of the sensory and motor blocks, and the most severe pain score during a 24-hour postoperative period.