Effective, scalable dementia prevention interventions are needed to address modifiable risk factors given global burden of dementia and challenges in developing disease-modifying treatments. A single-blind randomized controlled trial assessed an online multidomain lifestyle intervention to prevent cognitive decline over 3 years. Participants were dementia-free community-dwelling Australians aged 55-77 years with modifiable dementia risk factors. Eligible participants (n = 6,104, 64% female) were randomized 1:1 to a personalized schedule of online coaching in two to four modules (targeting physical activity, nutrition, cognitive activity and depression or anxiety) or a control group that received module-eligible information only. At 3 years, the mean change in a global cognitive composite, the primary outcome, was met. The mean changes in z scores were 0.28 (95% confidence interval (CI): 0.25-0.32) for intervention, 0.10 (95% CI: 0.07-0.13) for control and 0.18 (95% CI: 0.13-0.23, P <
0.001) for the between-group difference. Trial-related adverse events occurred in 19 (0.60%) intervention and 1 (0.03%) control participant. Randomization of this internet-delivered lifestyle intervention tailored to individual dementia risk factors resulted in significantly better cognition in older adults over 3 years. This intervention is scalable with the potential for population-level rollout that may delay cognitive decline in the general community. Australian New Zealand ClinicalTrials.gov registration: ACTRN12618000851268.