[en] Background and purposes. To define new directions, the Trans Tasman Radiation Oncology Group (TROG) has conducted a detailed analysis of its unrandomised experience with radio-chemotherapy in oesophageal cancer. Methods and patients. Since 1984, 373 patients with oesophageal cancer have been treated on three prospective, but unrandomised, protocols involving radiation with concurrent cisplatin and infusional fluorouracil. Centres in Australia and New Zealand have contributed patients. Reasons for case selection have been examined in detail and prognostic models have been examined in the light of biases exposed. Results. Cause specific survival in 92 patients treated pre-operatively with 35 Gy, infusional fluorouracil and cisplatin was 25.5 ± 6.0% at 5 years and similar to the 5 year expectations of 169 patients treated with 60 Gy and two courses of the same chemotherapy (23.8 ± 4.7%). Analysis of failure in these groups suggests that local relapse precedes the development of metastases and competes as a cause for ultimate failure. Although patients treated surgically were less likely to relapse locally, survival was no better because more developed metastases. Some of the 112 patients treated 'palliatively' with 30-35 Gy concurrent with chemotherapy without surgery have become long-term survivors with 5 year survival figure in this group 7.7 ± 3.4%. Apart from variables related to disease stage and performance status at presentation, tumour site emerged as a strong predictor of outcome. Prognosis worsens the nearer the tumour is to the stomach. In addition, indications of a radiation dose response relationship emerged. Conclusions. Concurrent radio-chemotherapy protocols can improve outcome in patients fit enough to tolerate these approaches. New strategies remain necessary, however