Abstract Objectives: vascular lesions are important contributors to cerebral disease, )ret relatively little work has been done on relationships between haematological factors and cognitive function. We have explored these relationships in the Caerphilly cohort of older men. Design: the development of vascular disease and the decline in cognitive function are being studied in a cohort of older men in Caerphilly, South Wales, UK. We have recorded a range of life-style, dietary, lipid, haematological and theological factors which are, or may be, predictive of vascular disease. We have examined the relationships between these and both incident myocardial infarction and stroke, and identified factors relevant to cognitive function. Setting: a community-based study based upon a representative population sample of older men. We collected data for the study from around 90% of the survivors of the original cohort. Subjects: 2154 men who were aged 55-69 years at the time blood was taken for the haematological tests and cognitive function was tested. Main outcome measures: we present data from the AH4 test of general cognitive performance and from choice reaction time, a test of vigilance. Results: there is a U-shaped relationship between haematocrit and cognitive function, the best cognitive function scores occurring at a mean haematocrit of 0.46. The relationship is significant for the choice reaction time (P < 0.05). The relationship between haematocrit and the AH4 test score is also curvilinear, but it is not significant. Plasma viscosity showed strong and robust relationships, with significantly better cognitive performance and faster reaction times at lower levels of viscosity (both relationships P < 0.001). Plasma fibrinogen concentration did not show any significant relationships with cognitive function. All these... [ABSTRACT FROM AUTHOR]