The presence of hypoxic cells in human tumors is thought to be one of the principal reasons for the failure of radiation therapy. Intensive laboratory and clinical efforts to overcome tumor hypoxia have focused on oxygenating, radiosensitizing or killing the maximally radioresistant fraction of tumor cells. This "hypoxic fraction" dominates the single-dose radiation response, irrespective of the oxygenation status of the remainder of the tumor cell population. However, at doses that are typical of those delivered in a daily radiotherapy protocol, we show that the tumor response is highly dependent upon the cells at oxygen levels intermediate between fully oxygenated and hypoxic (0.5-20 mm Hg). For most tumors, these cells are more important than the radiobiologically hypoxic cells in determining treatment outcome after 30 fractions of 2 Gy. We also show that under conditions of diffusion-limited hypoxia, the impact of full reoxygenation between fractions is much smaller than previously realized. Together, the results imply that tumor hypoxia plays a more significant role in determining the outcome of fractionated radiotherapy than previous measurements and assumptions of hypoxic fractions have indicated. Therefore, the concept of a hypoxic fraction in human tumors is less meaningful when pertaining to a fractionated radiotherapy regimen, and should not be expected to be useful for predicting tumor responses in the clinic. This implies the need to characterize tumor oxygenation in a manner that reflects the true oxygenation status of all the tumor cells, not just the ones most refractory to the effects of ionizing radiation. Furthermore, effective therapeutic agents must have the ability to specifically sensitize or kill those cells at intermediate levels of oxygen in addition to the radiobiologically hypoxic cells.