In order to define clinically relevant lactic acidosis, 12 biochemical variables, eight clinical symptoms and signs, leading diagnoses, and mortality were evaluated prospectively in approximately 2,000 unselected patients with internal diseases, consecutively admitted to the hospital. Patients with incomplete data sets were not considered. Of those patients who repeatedly were admitted to the hospital during the time of the study, only the first admission was included for statistical analysis. In addition to 11 definitions of lactic acidosis given in the literature, sequential cluster analyses of the biochemical variables were used to estimate the incidence of lactic acidosis in 1,467 patients. Depending upon which definition was used, 0.5-3.8% of all patients were classified as suffering from lactic acidosis, with a mortality rate ranging from 30-88%. From this study it is concluded that a limit of less than or equal to 7.35 for pH and of greater than 5-6 mmol/L for the concentration of lactate in whole blood will minimize false-negative or false-positive classifications.