Blood pressure was assessed between 1981 and 1983 in all persons over age 65 years in three communities (East Boston, Massachusetts; New Haven, Connecticut; and Iowa and Washington counties, Iowa), and cause-specific mortality was monitored annually over the subsequent 5 years as part of the National Institute on Agingsponsored Established Populations for Epidemiologic Studies of the Elderly. Each community had 80% or more participation: in East Boston, 3,809 persons with 903 deaths, in New Haven, 2,812 persons with 804 deaths, and in Iowa, 3,673 persons with 763 deaths. At 2 years, odds of death from all causes were higher in the low (<130 mmHg) than the middle (130–159 mmHg) systolic blood pressure group for persons aged 65–79 years in all three populations. By 5 years, cardiovascular death increased with increasing systolic pressure in all three communities and reached significance in Iowa. Cancer death was highest in the low systolic pressure stratum in all three centers. All-cause, cardiovascular death, and cancer mortality was highest in the low (<75 mmHg) diastolic blood pressure group in East Boston, even at 5 years. Blood pressures obtained 9 years earlier in 2, 079 (68%) of the East Boston participants showed a significantly higher risk of cardiovascular death with increasing systolic pressure and no relation between diastolic pressure and mortality risk. In the elderly, excess mortality at lower levels of blood pressure during early follow-up may in part be due to the effects of illness and disability present at baseline. This may obscure the usual rise in mortality with increasing systolic pressure. There is no consistent relation between diastolic pressure and mortality.