Three cohorts of men aged 40 to 59 at entry, for a total of 2480 subjects, were examined for the measurement of some cardiovascular risk factors and then followed-up for 20 years. Changes of systolic blood pressure (SBP) over the first 10 years in 2088 men and computed by an integral-like procedure (delta-SBP) were used as possible predictors of fatal events recorded in the second 10 years of follow-up. Men who had a relative increase of systolic blood pressure in the first 10 years showed an excess risk of death as compared to those who had a relative decrease of systolic blood pressure (adjusted for age and for SBP at entry levels). The relative risk between the two groups was 1.38 for all causes of mortality (ALL), 1.50 for the pool of atherosclerotic cardiovascular disease mortality (ACVD), 1.68 for coronary heart disease deaths (CHD) and 1.19 for stroke deaths (STR). These relative risks are all statistically significant except for strokes. The Cox model was solved using the same four end-points as dependent variables (ALL, ACVD, CHD, STR, separately) and as covariates some risk factors measured at entry examination (age, cigarettes smoked per day, body mass index, serum cholesterol, physical activity at work, systolic blood pressure) and, in addition, the indicator of SBP changes (delta-SBP). By adding delta-SBP the predictive power of the model was significantly improved and the coefficient of delta-SBP proved to be statistically significant (P less than .05). This was true for all end points except stroke.(ABSTRACT TRUNCATED AT 250 WORDS)