The purpose of our study was to estimate the extent to which casual blood pressure measurements can predict mortality in elderly people in a defined population. A random sample of subjects born between 1890 and 1914 was extracted and their medical records were studied. Their earliest blood pressure record was chosen, provided the patient was at least 60 years old at the time of measurement. The population consisted of 961 patients (49.8% women), 560 of whom died between 1968 and 1987. Their mean age was 70.1 years and the total number of person-years involved was 8625. The mortality risk was estimated as a function of sex, present age, systolic (SBP) and diastolic (DBP) blood pressures, by a version of the Cox regression model. On the basis of close confidence intervals we conclude that blood pressure in the elderly is a very weak predictor of mortality.