BACKGROUND: Some cohort studies have reported a decline in dementia prevalence and incidence over time, although these findings have not been consistent across studies. We reviewed evidence on changes in dementia prevalence and incidence over time using published population-based cohort studies that had used consistent methods with each wave and aimed to quantify associated changes in risk factors over time using population attributable fractions (PAFs). METHODS: We searched for systematic reviews of cohort studies examining changes in dementia prevalence or incidence over time. We searched PubMed for publications from database inception up to Jan 12, 2023, using the search terms "systematic review" AND "dementia" AND ("prevalence" OR "incidence"), with no language restrictions. We repeated this search on March 28, 2024. From eligible systematic reviews, we searched the references and selected peer-reviewed publications about cohort studies where dementia prevalence or incidence was measured in the same geographical location, at a minimum of two timepoints, and that reported age-standardised prevalence or incidence of dementia. Additionally, data had to be from population-based samples, in which participants' cognitive status was assessed and where validated criteria were used to diagnose dementia. We extracted summary-level data from each paper about dementia risk factors, contacting authors when such data were not available in the published paper, and calculated PAFs for each risk factor at all available timepoints. Where possible, we linked changes in dementia prevalence or incidence with changes in the prevalence of risk factors. FINDINGS: We identified 1925 records in our initial search, of which five eligible systematic reviews were identified. Within these systematic reviews, we identified 71 potentially eligible primary papers, of which 27 were included in our analysis. 13 (48%) of 27 primary papers reported change in prevalence of dementia, ten (37%) reported change in incidence of dementia, and four (15%) reported change in both incidence and prevalence of dementia. Studies reporting change in dementia incidence over time in Europe (n=5) and the USA (n=5) consistently reported a declining incidence in dementia. One study from Japan reported an increase in dementia prevalence and incidence and a stable incidence was reported in one study from Nigeria. Overall, across studies, the PAFs for less education or smoking, or both, generally declined over time, whereas PAFs for obesity, hypertension, and diabetes generally increased. The decrease in PAFs for less education and smoking was associated with a decline in the incidence of dementia in the Framingham study (Framingham, MA, USA, 1997-2013), the only study with sufficient data to allow analysis. INTERPRETATION: Our findings suggest that lifestyle interventions such as compulsory education and reducing rates of smoking through country-level policy changes could be associated with an observed reduction, and therefore future reduction, in the incidence of dementia. More studies are needed in low-income and middle-income countries, where the burden of dementia is highest, and continues to increase. FUNDING: National Institute for Health and Care Research Three Schools' Dementia Research Programme.