There is substantial evidence for the presence of hypoxia in human tumours. This is documented by histopathological demonstration of vascular insufficiency, direct oxygen measurements in tumours, as well as by physiological imaging and mapping of hypoxic areas. As a consequence, clinical trials have focused on the hypoxia problem for more than 30 years. This includes the use of hyperbaric oxygen, hypoxic cell radiosensitizers, and, more recently, modification of the oxygen-unloading capacity of haemoglobin. Agents directed towards destruction of hypoxic cells have also been applied, such as hyperthermia and bio-reductive drugs. Despite decades of clinical trials, the results are still inconclusive, and although some trials have shown significant benefit, it has become apparent that hypoxia is a complex problem. Hypoxia appears to be especially a problem in certain tumour types (e.g. squamous cell carcinoma), but even within tumours of the same type, site, and stage, hypoxia does not occur to the same extent. Furthermore, there are increasing suggestions that hypoxia may occur in two principally different ways, namely acutely and chronically, yielding varying responses to modifying agents. Although improvement in hypoxic cell radiosensitizers and other agents is under way, a definitive solution to the hypoxia problem will not be found until the tumours in which hypoxia occurs can be identified. This will require detailed analysis of individual tumours and patients' parameters, and better knowledge of the mechanisms of reoxygenation in clonogenic tumour cells.