The aim of this work was to evaluate whether pulse pressure (PP) in elderly people is a better predictor of coronary mortality than systolic and diastolic blood pressure taken alone. For this aim, 3282 elderly subjects aged у65 years were studied in a population-based frame. Blood pressure was repeatedly measured and averaged; historical data, anthropometrics, blood tests and 14-year coronary mortality were recorded. Statistics included analysis of covariance, Cox analysis and bivariate vectorial analysis. Coronary mortality in women was predicted by PP (1.01 excess risk/mm Hg PP) and was significantly higher in the 3rd than in the 1st tertile of PP (relative risk 2.90); neither systolic nor diastolic pressure taken alone influenced mortality. When systolic and diastolic pressures were both entered into a Cox model, the former had a positive and