Within this special issue, many eminent investigators report on measurements of oxygen (O2) levels in tissues. Given the complexities of spatial and temporal heterogeneities of O2 in tissues and its many sources, this commentary draws attention to what such measurements do and do not actually assess regarding O2 levels in tissues. Given this limitation, it also discusses how these results can be used most effectively. To provide a convenient mechanism to discuss these issues more fully, this analysis focuses on measurements using EPR oximetry, but these considerations apply to all other techniques. The nature of the delivery of O2 to tissues and the mechanisms by which O2 is consumed necessarily result in very different levels of O2 within the volume of each voxel of a measurement. Better spatial resolution cannot fully resolve the problem because the variations include O2 gradients within each cell. Improved resolution of the time-dependent variation in O2 is also very challenging because O2 levels within tissues can have fluctuations of O2 levels in the range of milliseconds, while most methods require longer times to acquire the data from each voxel. Based on these issues, we argue that the values obtained inevitably are complex aggregates of averages of O2 levels across space and time in the tissue. These complexities arise from the complex physiology of tissues and are compounded by the limitations of the technique and its ability to acquire data. However, one often can obtain very meaningful and useful results if these complexities and limitations are taken into account. We illustrate this, using results obtained with in vivo EPR oximetry, especially utilizing its capacity to make repeated measurements to follow changes in O2 levels that occur with interventions and/or over time.