The aim of the CASTEL, a population-based (n ؍ 3282) prospective study which began 14 years ago, was to identify those items which had a prognostic impact in the elderly, and to evaluate whether the typical cardiovascular risk factors, particularly arterial hypertension, play a role after the age of 65 years. Initial screening, final follow-up and annual detection of mortality were performed. Mantel-Hanszel approach and multivariate Cox model were used for statistics. Cardiovascular mortality was 23.3% in normotensive, 23.3% in borderline, and 25% in the sustained hypertensive subjects (insignificant difference). In women, the incidence of stroke and coronary artery disease weakly depended on pulse pressure. Historical stroke and myocardial infarction predicted cardiovascular mortality in women; diabetes, uricaemia and high heart rate in men. In the very old, the predictors were less numerous, and blood pressure was not a predictor whatsoever; pulse blood