INTRODUCTION: previous studies have suggested that smoking, living alone and having a high body mass index may increase risk of developing dementia whereas a normal body mass index, having received education and moderate alcohol consumption may decrease risk. Dementia risk also increases with age and is thought to be higher in hypertensives. METHOD: we used data collected in the Hypertension in the Very Elderly Trial (HYVET), and cognitive function was assessed using the Mini-Mental State Examination (MMSE) at baseline and annually. Participants with a fall in MMSE to <24 or with a fall of 3 points in any 1 year were investigated further. The association of baseline sociodemographic, medical and lifestyle factors with incident dementia or decline in MMSE scores was assessed by regression models. RESULTS: incident dementia occurred in 263 of 3,336 participants over a mean follow-up of 2 years. In multivariate analyses, being underweight, BMI < 18.5 (HR 1.90, 95% CI 1.06-3.39) or obese, BMI >30 (HR 1.84, 95% CI 1.24-2.72), increased risk of incident dementia as did piracetam use (HR 2.72, 95% CI 1.60-4.63). Receiving formal education was associated with a reduced risk (HR 0.59, 95% CI 0.45-0.78). There was no association with smoking, alcohol and gender. Similar results were found when examining mean annual change in the MMSE score. DISCUSSION: our results for BMI and education agree with those from other studies. The increased risk associated with piracetam may reflect awareness of memory problems before any diagnosis of dementia has been made. Trial participants may be healthier than the general population and further studies in the general population are required.