Opioids are increasingly used in the treatment of chronic non-malignant pain. The aim of this open-label, randomised, parallel group study was to compare analgesia and side-effects of two commonly used opioid analgesics, tramadol and dihydrocodeine, in long-acting formulations in 60 osteoarthritis patients with strong pain despite NSAID's. Dose titration based on effect was performed with the respective immediate release solutions given additionally to tramadol 100 mg bid and dihydrocodeine 60 mg bid during the first 4 days of the 1 month treatment. Electrical sensation and pain thresholds over the osteoarthritic joint and at a distant location and gastrointestinal transit times were performed before and during treatment. Thirty patients with pain controlled by NSAID's alone formed the comparator group. Pain intensities at rest and during movement decreased highly significantly with tramadol and dihydrocodeine from median pre-treatment verbal ratings of over 3 (0=none, 4=unbearable) to 1 and below from the second treatment day onwards (ANOVA P<0.0001). Pain at rest was significantly lower with tramadol (ANOVA P=0.04), but ratings were similar during movement. Mean (95% CI) daily doses on days 1 and 28 were 209 (198-220) mg and 203 (191-206) mg of tramadol, and 129 (122-136) mg and 130 (121-134) mg of dihydrocodeine, respectively. Minor side-effects were more common with tramadol (P=0.04). Changes in bowel functions and symptoms were minor with both treatments, but the frequency of defaecation was lower and stools were harder with dihydrocodeine. Orocaecal transit time remained unchanged and similar to controls with both analgesics. Colonic transit times only increased significantly during treatment with dihydrocodeine. Sensation and pain thresholds were lower pre-treatment in both groups than in controls and increased during treatment. These antinociceptive effects were more marked in the tramadol group and distant from the osteoarthritic joint. We conclude rapid pain relief was achieved with both long-acting tramadol and dihydrocodeine with NSAID's in strong osteoarthritis pain. Minimal dose titration was required and side-effects were minor. Tramadol interfered less with intestinal function and showed greater antinociceptive action.