By enabling noninvasive measurements of tissue biochemistry, nuclear magnetic resonance spectroscopy (MRS) provides a unique means of characterizing tissues. Differences in equipment, techniques, and methodology between different laboratories cause major difficulties when comparing results, whether from measurements of tissue metabolism, or from the effects of different therapies. This is of concern in critically evaluating work from different laboratories and centres, causing potential difficulties in reproducing results, limiting the establishment of MRS as a standard method of diagnosis and of characterising disease and response to therapy in the laboratory and clinic. It also poses particular problems in establishing the multicentre clinical trials of MRS that are now required to provide adequate statistical power in confirming the encouraging preliminary clinical observations. These difficulties arise principally from imperfect localization of signal from selected regions of interest in the body, and from the subsequent analyses of the MRS spectra. Improvement is possible by establishing agreed procedures for test measurements and for data analysis, and by using appropriate test objects and test substances to establish the quality of measurements. A concerted research project on characterisation of biological tissues by NMR, principally concerned with MR imaging (MRI), was activated in 1984 by the European Economic Community as part of its third Medical and Health Research Programme, under the auspices of the Biomedical Engineering Concerted Actions' Committee (COMAC-BME). In 1988, this project was prolonged for 5 years, when the programme was expanded to encompass MRS. A series of five accompanying papers describes (a) the test protocols and objects agreed for assessing the quality of volume selective MRS measurements; (b) the experimental trials performed for a multicentre evaluation of these procedures on experimental and clinical systems; and (c) the results of a joint quantitative data analysis study on in vivo NMR time-domain test signals.