Clinical case-control studies of the relation between blood pressure and cognitive function have generally found lower function among hypertensives. Most of these studies were small and incompletely controlled for confounders. Two population-based studies have yielded conflicting results. This study examines cognitive function over the entire range of blood pressure in a defined elderly population. A questionnaire administered in the home to 3,809 persons aged greater than or equal to 65 years in East Boston, Massachusetts, in 1982 and 1983 contained four brief cognitive tests: immediate memory, delayed memory, a mental status questionnaire, and digit span. In linear regression analyses adjusting for age, sex, and education, the direction of the association was not consistent among the tests. An increase in diastolic pressure of 10 mmHg was associated with an increase of 1.0 in percentile scores on the immediate memory test (95% confidence interval (CI) -0.04 to 1.9); with an increase of 1.1 in percentile scores on the delayed memory test (95% CI -0.1 to 2.3); with a decrease of -0.8 in percentile scores on the mental status questionnaire (95% CI -1.8 to 0.2); and with a decrease of -0.9 in percentile scores on the attention test (95% CI -1.5 to -0.2). These results suggest that blood pressure is not a substantial contributor to cognitive status in the elderly.