Combined modality treatment with radiotherapy and chemotherapy is used increasingly for the primary management of a variety of human tumours, with the aim of improving both local and distant control. The present paper reviews methodological issues related to the evaluation of combined modality therapy. Reports that patients have superior outcome in single-arm studies as compared to historical controls treated with radiation alone have limited value because of several types of bias including patient selection, stage migration, the tendency to publish positive results, or inadequate follow-up as compared to the historical series. The observation that response to chemotherapy predicts for survival after combined treatment also conveys no proof that combined treatment is superior to radiation alone. Randomized controlled trials provide the only rigorous method for evaluating combined therapy, but are also subject to misinterpretation. The majority of published trials report negative results but are too small to detect clinically important differences in survival. Even large trials may give spurious results if they seek small benefits of treatment in a spectrum of patients with widely differing prognosis. Some randomized trials have demonstrated improved local control and increased toxicity from combined treatment, a result that might have been achieved by increasing the effective radiation dose. Ideally, combined treatment should be compared with radiotherapy alone at equal levels of normal tissue damage. A review of published data for patients with cancers of the head and neck, lung, gastrointestinal tract and bladder reveals very few trials which have adequately evaluated the role of combined modality therapy (with or without surgery). Most of the large randomized trials have demonstrated no benefit from the use of radiation and chemotherapy, although some of them suggest small therapeutic gains from using thoracic radiation with chemotherapy in small-cell-lung cancer of limited extent, or from combined modality treatment after resection of rectal cancer. Possible reasons for the failure of active drugs to lead to easily detected gains in therapeutic index include insufficient reduction in cell survival from chemotherapy, selective killing of radiosensitive subpopulations, stimulation of the proliferation of surviving cells, or enhancement of metastasis. With the possible exception of radiation and concurrent 5-fluorouracil for squamous cancers of the anal canal, there are no convincing data to mandate the routine combined use of radiotherapy and chemotherapy in any of the above sites.