Background:Dementia is a leading cause of death in older adults in the United States, and the population at greatest risk is expanding as post-war birth cohorts move through later life. Existing national projections have not consistently resolved the demographic and geographic axes along which older adults experience dementia. We examined age-adjusted dementia mortality from 1999 to 2025 and projected mortality through 2050, overall and across strata relevant to aging health and social care planning. Methods:We conducted an ecological time-series analysis using the CDC Wide-ranging Online Data for Epidemiologic Research (CDC WONDER) Multiple Cause of Death database. Dementia deaths were identified using the International Classification of Diseases, Tenth Revision codes F01, F03, and G30–G31, listed anywhere on the death certificate. Annual age-adjusted mortality rates (AAMR) per 100,000 population were calculated using the 2000 U.S. Standard Population. Temporal trends were modelled with joinpoint regression to estimate the annual percent change (APC) and average annual percent change (AAPC) with 95% confidence intervals (CI). Mortality was projected through 2050 using a demographically optimized ensemble model in Python, with subgroup-specific calibration and capped weighting applied where mathematically derived weights were unstable. The forecast error was assessed using the mean error andmean absolute percentage error. Analyses were conducted for the overall population and stratified by sex, age, race and ethnicity, urbanization, census region, and state. Adults aged ≥65 years were the primary geriatric stratum, with adults aged 25–64 years reported as context. Results:Between 1999 and 2025, 9,046,693 dementia-related deaths were documented. The overall AAMR rose from 86.83 to 166.00 (AAPC 2.52; 95% CI 1.00 to 4.06; p=0.001) and is projected to reach 207.28 by 2050. Adults aged ≥65 years carried nearly all the burden (mean AAMR 735.98) and are projected to reach 1,022.42 by 2050. Women had consistently higher rates than men (mean 151.89 vs. 133.67), and the absolute female–male gap is projected to widen through 2050. Projections across racial and ethnic groups diverged: AAMRs are projected to continue rising in non-Hispanic White (+30.1%), non-Hispanic Asian or Pacific Islander (+74.6%), and non-Hispanic American Indian or Alaska Native (+35.2%) adults but to decline in non-Hispanic Black (–35.2%) and Hispanic or Latino adults (–17.1%). Non-metropolitan AAMRs exceeded metropolitan AAMRs from 1999 to 2020 (147.61 vs. 139.46); however, projections indicate that metropolitan rates will surpass non-metropolitan rates by 2050. Regional projections also diverged, with declines in the South contrasting with continued increases in the Midwest, Northeast, and West. Over half of dementia deaths (51.90%) occurred in nursing homes or long-term care settings, and 21.05% occurred at home. Conclusions:U.S. dementia mortality has risen substantially since 1999 and is projected to continue rising overall through 2050, with divergent trajectories across racial and ethnic groups, urbanization categories and regions. Projections for low-frequency strata, including non-Hispanic American Indian or Alaska Native and Hispanic or Latino adults, carry wider uncertainty and should be interpreted directionally rather than precisely. The concentration of dementia deaths in long-term care and home settings, combined with shifting inequalities, argues for parallel investment in equitable diagnosis, caregiver support, and long-term care capacity, aligned with the demographic geography of older adults.