We compared the relations of 4 blood pressure (BP) indexes (pulse pressure [PP], systolic BP [SBP], diastolic BP [DBP], and mean arterial pressure [MAP]) with 25-year mortality rates for coronary heart disease (CHD), cardiovascular disease (CVD), and all causes in younger, middle-aged, and older men and women by using data from a long-term prospective epidemiological study of employed persons who were screened between 1967 and 1973. A single supine BP measurement was obtained at baseline. Vital status was determined through 1995. We report on 5 groups (total, 28 360 participants) consisting of men age 18 to 39, 40 to 59, and 60 to 74 years and of women age 40 to 59 and 60 to 74 years who were not receiving antihypertensive treatment, had no history of CHD, and did not have diabetes. Cox proportional hazards analyses were used to determine multivariate-adjusted hazard ratios with a 1-SD higher value for each BP index; Wald chi(2) tests were used to compare the strength of relations. Relations of PP were less strong than were those of SBP for all end points in all age/gender groups. SBP or MAP showed the strongest relations to all end points in all age/gender groups (hazard ratio, 1.17 to 1.36). The relations of SBP to death were stronger than were those of DBP, except for middle-aged men and for CVD in women. DBP showed significant positive associations with death, after control for SBP, in middle-aged participants. In conclusion, these data indicate that the long-term risk of high BP should be assessed mainly on the basis of SBP or of SBP and DBP together, not on the basis of PP, in apparently healthy adults.