Several studies have shown that low blood pressure in individuals age 65 years and older is related to increased overall mortality. We hypothesize that this association is secondary to serious underlying illness, which has caused blood pressure reduction and, subsequently, has increased the risk of dying. Our study population was comprised of individuals age 20 years and older in the county of Nord Trøndelag in Norway, who were studied in a general health survey between 1984 and 1986. We had measurements of blood pressure, blood glucose, weight, height, and other information for 9,732 women and 8,290 men age 65 years or older. During approximately 6 years of follow-up, 2,122 women and 2,578 men died. For both genders, low systolic pressure was not associated with increased mortality, and the mortality curve did not display a J-shaped relation, after adjustment for age, marital status, body mass index, blood glucose, self-assessed health, use of antihypertensive medication, and history of diabetes and cardiovascular diseases. For diastolic pressure, however, women in the lowest category (< 75 mmHg) had an adjusted mortality rate ratio of 1.21 (95% confidence limits = 1.05, 1.39), compared with reference women (80-87 mmHg). Among men, the analogous mortality rate ratio was 1.16 (95% confidence limits = 1.02, 1.31). To reduce further the potential confounding between diastolic pressure and underlying illness, we excluded users of antihypertensive medication as well as the 2 first years of follow-up. After these procedures, the J-shaped mortality curve was not present among women, and it was substantially reduced among men. Thus, the results for both men and women indicated that the age-adjusted J-shaped relation between diastolic blood pressure and mortality was confounded with indicators of ill health, and that the often-found association between low diastolic blood pressure and increased mortality is indirect, possibly caused by serious underlying disease.