A community cohort of 1270 non-demented 75+ years old persons was followed to evaluate the influence of blood pressure on incidence of dementia. Two hundred and eighteen dementia cases were detected during an average of three years of follow-up. Subjects with baseline systolic pressure >/= 180 mm Hg had an age- and gender-adjusted relative risk (RR) of 1.6 (95% to persons with systolic pressure of 141-160 mm Hg. This association persisted, although not statistically significant, when education, vascular diseases, and antihypertensive drug use were entered in the model (RR = 1.4; 95% systolic pressure were not related to dementia incidence. However, individuals with a decrease of 5--19 mm Hg and >/= 20 mm Hg in systolic pressure from baseline to follow-up had a RR of 1.8 (95% 2.6) and 2.5 (95% also found in subjects with diastolic pressure reduction. In conclusion, our findings support an association between high systolic pressure and increased risk of dementia, whereas blood pressure reduction may be secondary to the dementia process itself.