Clinical signs of hypoxia and hyperoxia are nonspecific and unreliable, yet both are potentially injurious. Noninvasive methods of oxygen assessment fill the gap between clinical observation and invasive tests, helping physicians deliver sufficient oxygen with minimum toxicity. Potential sites for oxygen measurement vary between the blood and the mitochondria; each method measures at a different site and detects different types of hypoxia and hyperoxia. Thus, values obtained by two different methods are not equivalent, giving each method unique strengths and weaknesses. We review two clinical methods (pulse oximetry and transcutaneous oximetry), as well as four experimental methods (near-infrared spectrophotometry, magnetic resonance spectroscopy, magnetic resonance saturation imaging, and time-of-flight absorbance spectrophotometry). The principles of each method and the clinical situations in which each succeeds or fails are discussed. A fundamental understanding of each method can help in deciding which methods, if any, are appropriate for a given patient and how best to correct observed oxygenation problems once they are discovered.