BACKGROUND: Dementia is a leading, global public health challenge. Recent evidence supporting a decrease in age-specific incidence of dementia in high-income countries (HICs) suggests that risk reduction is possible through improved life-course public health. Despite this, efforts to date have been heavily focused on individual-level approaches, which are unlikely to significantly reduce dementia prevalence or inequalities in dementia. In order to inform policy, we identified the population-level interventions for dementia risk reduction with the strongest evidence base. METHODS: We did this complex, multistage, evidence review to summarise the empirical, interventional evidence for population-level interventions to reduce or control each of the 12 modifiable life-course risk factors for dementia identified by the Lancet commission. We conducted a series of structured searches of peer-reviewed and grey literature databases (eg, Medline, Trip database, Cochrane library, Campbell Collaboration, the WHO, and Google Scholar), in January, March, and June, 2023. Search terms related to risk factors, prevention, and population-level interventions, without language restrictions. We extracted evidence of effectiveness and key contextual information to aid consideration and implementation of interventions by policymakers. We performed a narrative synthesis and evidence grading, and we derived a population-level dementia risk reduction intervention framework, structured by intervention type. This study is registered with PROSPERO, ID:CRD42023396193. FINDINGS: We identified clear and consistent evidence for the effectiveness of 26 population-level interventions to reduce the prevalence of nine of the risk factors, of which 23 have been empirically evaluated in HICs, and 16 in low-income and middle-income countries. We identified interventions that acted through fiscal levers (n=5; eg, removing primary school fees), marketing or advertising levers (n=5; eg, plain packaging of tobacco products), availability levers (n=8; eg, cleaner fuel replacement programmes for cooking stoves), and legislative levers (n=8; eg, mandated provision of hearing protective equipment at noisy workplaces). We were not able to recommend any interventions for diabetes (other than indirectly through action on obesity and physical inactivity), depression, or social isolation. INTERPRETATION: This complex evidence review provides policymakers and public health professionals with an evidence-based framework to help develop and implement population-level approaches for dementia risk reduction that could significantly reduce the population's risk of dementia and reduce health inequalities. FUNDING: None.