A retrospective study of the medical records of 115 consecutive cases of meningococcal disease with 11 fatalities was carried out in order to evaluate the power of 15 clinical and laboratory variables available on admission in predicting a fatal outcome. On linear discriminant analysis, six variables showed a significant discriminating power in predicting death: low systolic blood pressure, low platelet count, extensive petechiae, high body temperature, low CSF polynuclear cell count and absence of meningism. From a stepwise linear discriminant analysis, two alternative procedures for prognostic evaluation were derived. If a large high risk group is accepted which will include practically all patients at risk of death, a prognostic evaluation based only on systolic blood pressure on admission is sufficient. Alternatively, if unconventional, potentially hazardous therapy is considered for high risk patients, a small high risk group may be defined. Patients with systolic blood pressure less than 100 mm Hg, platelet count less than or equal to 125 X 10(9)/l and body temperature greater than 39 degrees on admission constitute a small group with a very high mortality. A similar risk group was defined if platelet count less than or equal to 125 X 10(9)/l was substituted for extensive petechiae. The advantage of the latter procedure is that only a simple bedside examination is required for the prognostic evaluation.