BACKGROUND: Cardiovascular risk factors and a history of vascular disease can increase the risk of Alzheimer's disease (AD). AD is less common in aspirin users than non-users, and there are plausible biological mechanisms whereby aspirin might slow the progression of either vascular or Alzheimer-type pathology. We assessed the benefits of aspirin in patients with AD. METHODS: 310 community-resident patients who had AD and who had no potential indication or definite contraindication for aspirin were randomly assigned to receive open-label aspirin (n=156; one 75-mg enteric-coated tablet per day, to continue indefinitely) or to avoid aspirin (n=154). Primary outcome measures were cognition (assessed with the mini-mental state examination [MMSE]) and functional ability (assessed with the Bristol activities of daily living scale [BADLS]). Secondary outcomes were time to formal domiciliary or institutional care, progress of disability, behavioural symptoms, caregiver wellbeing, and care time. Patients were assessed at 12-week intervals in the first year and once each year thereafter. Analysis of the primary outcome measures was by intention to treat. This study is registered as an International Standard Randomised Controlled Trial, number ISRCTN96337233. FINDINGS: Patients had a median age of 75 years; 156 patients had mild AD, 154 had moderate AD, and 18 had concomitant vascular dementia. Over the 3 years after randomisation, in patients who took aspirin, mean MMSE score was 0.10 points higher (95% CI -0.37 to 0.57; p=0.7) and mean BADLS score was 0.62 points lower (-1.37 to 0.13; p=0.11) than in patients assigned to aspirin avoidance. There were no obvious differences between the groups in any other outcome measurements. 13 (8%) patients on aspirin and two (1%) patients in the control group had bleeds that led to admission to hospital (relative risk=4.4, 95% CI 1.5-12.8; p=0.007); three (2%) patients in the aspirin group had fatal cerebral bleeds. INTERPRETATION: Although aspirin is commonly used in dementia, in patients with typical AD 2 years of treatment with low-dose aspirin has no worthwhile benefit and increases the risk of serious bleeds.