Tumor hypoxia is of considerable importance to the oncologist in selecting and optimizing cancer therapy, because hypoxia can determine the effectiveness of various therapies. The relationship between tumor hypoxia and tumor bioenergetics, assessed by 31P MR spectroscopy, is examined to determine whether 31P MR spectroscopy can be clinically useful to measure or characterize tumor hypoxia. Work with experimental tumors has suggested that several different types of hypoxia may exist and that 31P MR spectroscopy cannot be used to characterize all types. Metabolic hypoxia is the level of hypoxia that results in mitochondrial impairment in cells, and it is associated with declining cellular bioenergetic status, which can be measured by enzymatic assay of adenosine triphosphate (ATP). Because 31P MR spectroscopy is sensitive to levels of ATP, it is potentially sensitive to metabolic hypoxia in vivo and may provide a rapid and noninvasive technique for characterizing metabolic hypoxia in tumors. Radiobiologic hypoxia is the level of hypoxia that results in attenuated cell death due to radiation, because radiotoxicity is directly related to tissue levels of oxygen. Radiobiologic hypoxia of tumors thus has more impact on choice of therapy, yet the relationship between metabolic hypoxia and radiobiologic hypoxia remains to be elucidated. An analysis of published data suggests that 31P MR spectroscopy is directly sensitive to metabolic hypoxia in tumors, but it is only indirectly sensitive to radiobiologic hypoxia in tumors. Therefore, 31P MR spectroscopy may be unable to quantify the cell fraction of a tumor that has radiobiologic hypoxia. However, preliminary data suggest that MR spectroscopy may prove useful for determining the effectiveness of therapeutic interventions designed to manipulate radiobiologic hypoxia in tumors or for monitoring the kinetics of tumor reoxygenation after treatment.