BACKGROUND AND PURPOSE: Despite numerous randomized trials investigating radiotherapy (RT) fractionation schedules for painful bone metastases, there are very few data on RT for bone metastases causing pain with a neuropathic component. The Trans-Tasman Radiation Oncology Group undertook a randomized trial comparing the efficacy of a single 8 Gy (8/1) with 20 Gy in 5 fractions (20/5) for this type of pain. MATERIALS AND METHODS: Eligible patients had radiological evidence of bone metastases from a known malignancy with no change in systemic therapy within 6 weeks before or anticipated within 4 weeks after RT, no other metastases along the distribution of the neuropathic pain and no clinical or radiological evidence of cord/cauda equina compression. All patients gave written informed consent. Primary endpoints were pain response within 2 months of commencement of RT and time to treatment failure (TTF). The hypothesis was that 8/1 is at least as effective as 20/5 and the planned sample size was 270 patients. RESULTS: Between February 1996 and December 2002, 272 patients were randomized (8/1:20/5=137:135) from 15 centres (Australia 11, New Zealand 3, UK 1). The commonest primary cancers were lung (31%), prostate (29%) and breast (8%); index sites were spine (89%), rib (9%), other (2%); 72% of patients were males and the median age was 67 (range 29-89). The median overall survival (95% CI) for all randomized patients was 4.8 mo (4.2-5.7 mo). The intention-to-treat overall response rates (95% CI) for 8/1 vs 20/5 were 53% (45-62%) vs 61% (53-70%), P=0.18. Corresponding figures for complete response were 26% (18-34%) vs 27% (19-35%), P=0.89. The estimated median TTFs (95% CI) were 2.4 mo (2.0-3.3 mo) vs 3.7 mo (3.1-5.9 mo) respectively. The hazard ratio (95% CI) for the comparison of TTF curves was 1.35 (0.99-1.85), log-rank P=0.056. There were no statistically significant differences in the rates of re-treatment, cord compression or pathological fracture by arm. CONCLUSIONS: 8/1 was not shown to be as effective as 20/5, nor was it statistically significantly worse. Outcomes were generally poorer for 8/1, although the quantitative differences were relatively small.