among older adults at increased risk of cognitive decline. Methods: 170 community-dwelling older adults from the Perth Metropolitan area, who were free of dementia, but had subjective memory complaints or Mild Cognitive Impairment were randomly allocated to a 24-week home-based physical activity program or usual care. Participants in the intervention group received a modified behavioural intervention package based on social cognitive theory. They were encouraged to perform at least 150 minutes of moderate intensity PA per week. The primary outcome was change on the ADAS-cog score. Pedometers were used for an objective measure of physical activity. Results: 138 participants completed the trial. The study intervention resulted in 110 minutes more exercise related activity per week than usual care. Analysis of variance for repeated measures showed that older adults in the intervention group had improved on average 1.1 points on the ADAS-cog relative to controls over 18 months (p 0.009). The effect of exercise was apparent by 6 months and persisted at the 12 and 18-months assessments. Conclusions: To our knowledge this is the first RCT to confirm a benefit of exercise on cognitive function in older adults with memory complaints. Our findings widen the range of health benefits ascribed to PA in advancing age. Adoption of similar programs in the community could represent an affordable and readily available intervention to decrease the burden of cognitive impairment in older people.