264 patients experiencing pain related to osteoarthritis of the hip and/or knee were enrolled in a double-blind randomised multicentre study. Those who reported moderate or severe pain despite taking paracetamol 4g daily during a 1-week lead-in period were randomised to treatment with tramadol HCl 100mg 3 times daily or dextropropoxyphene napsylate 100mg 3 times daily for 2 weeks. At the end of the second week of therapy, 71.6 and 70.4% of tramadol-treated patients and 53.2 and 50.5% of dextropropoxyphene-treated patients had symptom improvement during daily activities (p = 0.01) and on walking (p = 0.006), respectively. A larger proportion of tramadol-treated patients also reported improvement of pain during sleep as compared with patients administered dextropropoxyphene (tramadol: 44.3%; dextropropoxyphene: 30.3%; p = 0.04). Numerically, tramadol was also more effective than dextropropoxyphene for improvement of pain relief as measured on a visual analogue scale, but the difference did not reach statistical significance (tramadol: 48, dextropropoxyphene: 42; p = 0.12). No potentially serious adverse events were reported. Tramadol was associated with a significantly higher number of reported adverse events; the most commonly reported events were nausea, dizziness and vomiting, and more patients taking tramadol were withdrawn due to adverse events.In conclusion, tramadol was more effective than dextropropoxyphene in treating osteoarthritic pain, but tramadol was more often associated with adverse events. However, considering the nonserious nature of the events and the high number of fatal overdoses reported in the literature with dextropropoxyphene, tramadol appears to be a better alternative.