In this multicentre trial tramadol and buprenorphine were compared for the treatment of neoplastic pain no longer responsive to non-steroidal antiinflammatory drugs. A total of 131 adults (86 M, 45F) were treated with tramadol (one 100-mg slow-release tablet every 8-12 h), or buprenorphine (one sublingual 0.2-mg tablet every 6-8 h). The trial was to continue for up to six months. Most patients started treatment with 2-3 tablets/day in both groups, and the mean treatment period was 58 days for tramadol and 51 for buprenorphine. Almost all dose changes needed were made in the first fortnight in both treatment groups, and the largest number of patients dropped out because of inadequate pain relief or progression of the underlying disease. The results achieved in the first two weeks persisted throughout the rest of the trial, and the investigator's assessments on each patient's clinical chart corresponded closely with those that patients made in their own daily diaries. In the four hours after the first dose both drugs virtually halved the severity of pain (measured using a visual analogue scale), and this relief lasted throughout treatment. By the end of the first week the proportion of patients with strong/unbearable pain in the tramadol group had fallen significantly (from 98.4% to 48.1%, p < 0.05), as compared to a drop from 92% to 66.7% for buprenorphine. The quality of sleep also tended to improve in the tramadol group, with the proportion of patients enjoying good or deep sleep rising from 37% to 50%, as compared to 33% to 40-44% with buprenorphine. Karnofsky's and Spitzer's indices reflecting the quality of life did not change in the tramadol group; in the buprenorphine group the Karnofsky index dropped slightly after a fortnight (p < 0.05 between treatments). In the first two months of the trial the number of patients with no/moderate pain rose continuously in the tramadol group (71% and 80% after one and two months); the rise was less marked in the buprenorphine group (number of patients with mild/moderate pain, 45% and 65%). In both the short term and in the longer term, it was found that the levels of efficacy and acceptability were always significantly better in the tramadol group than in the buprenorphine group. General and biological safety in both drugs was good. The most typical side-effects were those characteristic of opioids (nausea and/or vomiting, drowsiness). Adverse reactions were reported in 17 patients taking tramadol (25%) and in 16 taking buprenorphine (26%). There were six drop-outs in the first group (9%) and seven in the second (11%). Serious symptoms arose more frequently in the buprenorphine group (19% cf. 10%). No signs of dependence or tolerance were noted.