BACKGROUND Phosphorus nuclear magnetic resonance spectroscopy has been proposed as a method of studying the metabolism of the myocardium in patients. Little is known about 31P nuclear magnetic resonance spectroscopy of diseased human hearts. METHODS Two donor hearts meeting the requirements for heart transplantation and 11 diseased hearts were removed during a transplantation procedure and were studied in a horizontal 2.35 T superconducting magnet. Spectra were obtained at 0 degrees C about 30 minutes after the excision. The areas of the inorganic phosphate peak (Pi) and of the phosphocreatine peak (PCr) were summed and expressed as a ratio with respect to the area of the beta ATP peak. RESULTS The ratio (Pi + PCr)/beta ATP was found to be significantly lower in five hearts with a myocardial infarct (0.77 (0.18)) than in hearts with dilated cardiomyopathy (1.25 (0.29)) and in normal hearts (1.69 (0.11)). The area of the phosphodiester peak was expressed as a ratio with respect to the area of the beta ATP peak: no differences were found between the three groups. CONCLUSIONS These results suggest that the phosphocreatine concentration is lower in ischaemic heart disease than in dilated cardiomyopathy and that the phosphodiester peak is probably not useful in distinguishing between these two types of heart disease.