We surveyed 1,545 subjects--recruited into the UK Medical Research Council elderly hypertension treatment trial between 1982 and 1987--to detect incident cases of dementia, identifying 50 cases of dementia, including 31 cases of probable or possible Alzheimer's disease (AD). These we compared with 223 unimpaired, unmatched controls from the same population for exposure to familial, cardiovascular, educational, and geographic risk factors for dementia. Our study confirms the association of family history of dementia with dementia (odds ratio [OR] = 4.36) and AD (OR = 4.69), and of advanced age with dementia (OR = 2.81). Rural residence exerted a protective effect for dementia (OR = 0.21) and AD (OR = 0.28). We report near-significant associations between AD and dementia and several cardiovascular risk factors (ECG ischemia, systolic hypertension, and smoking) among subjects lacking a family history of dementia. We postulate the existence of a nonfamilial form of dementia transcending traditional categories of multi-infarct dementia and AD, more common among urban residents, and mediated through vascular pathology. Risk factors reported elsewhere but not confirmed in this study were advanced maternal age and winter season of birth.