Background In our aging population the burden of dementia is increasing, necessitating the urgent identification of treatable risk factors. Small cross-sectional studies demonstrate associations between nonvalvular atrial fibrillation (NVAF), silent cerebral infarction and decreased cognitive function, but there are few longitudinal studies in this area. This thesis reports the results of a prospective longitudinal cohort study of cognitive decline in people with recent-onset NVAF compared to controls. To inform the thesis, an extensive literature review was undertaken . This included searches on NVAF and cognitive decline, NVAF and silent infarction, epidemiology of NVAF, other risk factors for cognitive decline, epidemiology of cognitive decline and the neuropsychological tests included in the CAFE battery. Methods 362 people over 60, screenedi n primary care, underwent baseline assessmenitn cluding a battery of neuropsychological tests, repeated at 12 months (n=304). Cases (n=175) with recent-onset NVAF, were matched for age, sex and GP practice with controls in sinus rhythm. Data were compared using SPSS software (version 11) with both parametric and non-parametric analysis. Results Baseline characteristics, including cognitive function, were similar for cases and controls. There was wide variation between individuals in change in performance on the neuropsychological tests over 12 months, with some improving and some deteriorating for each sub-test. Cases (NVAF) significantly (p<0.05) deteriorated in four subtests measuring attention/ non-verbal memory, and significantly (p<0.05) improved in two subtests measuring verbal memory. Controls significantly (p<0.05) deteriorated and improved in the same sub-tests as cases, but significantly (p<0.05) deteriorated in another three subtests measuring attention/non-verbal memory, and significantly (p<0.05) improved in another six subtests. Treatment with warfarin or aspirin did not appear to be associated with change in cognitive status. Conclusions At baseline there was no significant difference in cognitive function between cases in NVAF and controls in sinus rhythm. At follow-up there was no consistent relationship between NVAF and cognitive decline over 12 months, nor any apparent effect of antithrombotic therapy. Explanations include true independence of NVAF and cognitive decline, or too short a follow-up period. An additional follow-up at 36 months is underway to explore this further.