The presence of radiation resistant cells in solid human tumors is believed to be a major reason why radiotherapy fails to eradicate some such neoplasms. The presence of unperfused regions containing hypoxic cells may also contribute to resistance to some chemotherapeutic agents. This paper reviews the evidence that radiation resistant hypoxic cells exist in solid tumors, the assumptions and results of the methods used to detect hypoxic cells, and the causes and nature of tumor hypoxia. Evidence that radiation resistant hypoxic cells exist in the vast majority of transplanted rodent tumors and xenografted human tumors is direct and convincing, but problems with the current methodology make quantitative statements about the magnitude of the hypoxic fractions problematic. Evidence that radiation resistant hypoxic cells exist in human tumors is considerably more indirect than the evidence for their existence in transplanted tumors, but it is convincing. However, evidence that hypoxic cells are a significant cause of local failure after optimal clinical radiotherapy or chemotherapy regimens is limited and less definitive. The nature and causes of tumor hypoxia are not definitively known. In particular, it is not certain whether hypoxia is a chronic or a transient state, whether hypoxic cells are proliferating or quiescent, or whether hypoxic cells have the same repair capacity as aerobic cells. A number of new methods for assessing hypoxia are reviewed. While there are still problems with all of the new techniques, some of them have the potential of allowing the assessment of hypoxia in individual human tumors.