Nuclear magnetic resonance spectroscopy has progressed far since the original description of the phenomenon (30,31) and now permits noninvasive and harmless measurements to be repeatedly made of tissue biochemistry. Currently, there is a paucity of NMR data on normal human metabolism and the interpretation of spectra recorded from diseased tissues must be circumspect. This is further complicated by the inability of NMR spectroscopy to render accurate quantitative measurements and all observations must be considered to be compatible with, rather than diagnostic of, specific diseases. Nevertheless, NMR is an important addition to the clinician's armamentarium and these relatively expensive instruments should be situated in regional referral centres. Undoubtedly the most significant advances in NMR will only occur following the successful combination of the imaging and spectroscopic techniques, since that will enable anatomical and metabolic data to be obtained from a single site without the need for biopsy. Nuclear magnetic resonance spectroscopy has much to offer and the practical hazards appear to be few. In the next few years we will undoubtedly see an improvement in the quality of the data obtainable an a growing diversity of clinical applications.