Limited research exists investigating the association between multidimensional poverty and dementia in low-and middle-income countries (LMICs). To investigate the association between multidimensional poverty and dementia among adults aged 50 years or older living in South Africa. This cross-sectional study was conducted in Soweto, Johannesburg, South Africa, between November 11, 2019, and February 28, 2020. Participants included 227 adults aged 50 years or older. Data analysis was concluded from August 1 to 30, 2021. Multidimensional poverty included 7 dimensions that are central to well-being (education, health, economic activity, living standards, social participation, fair treatment, and psychological well-being) and 11 indicators of deprivation within those dimensions (limited access to education; severe limitation of activity; difficulty functioning; unemployment; deprivation of access to running water, electricity, and a flush toilet; lack of involvement in community groups; discrimination; depression; and decreased self-esteem). The 8-item Interview to Differentiate Aging and Dementia (Assessing Dementia 8 [AD8]) and the Rowland Universal Dementia Assessment Scale (RUDAS) were used to assess dementia. Level and depth of poverty were compared between adults with no dementia and those with a score above the threshold for either the AD8 or the RUDAS, or for both the AD8 and the RUDAS, adjusting for gender, age group, and marital status. Correlation analyses assessed the overlap of dimensions of deprivation. Associations between dementia and multidimensional poverty were investigated using a multivariable logistic regression model. A total of 227 adults (146 women [64.3%]; mean [SD] age, 63.7 [0.5] years) were included in the study; 101 (44.5%) had dementia identified by the AD8, 14 (6.2%) had dementia identified by the RUDAS, and 50 (22.0%) had dementia identified by both the AD8 and the RUDAS. More men than women did not have dementia (26 of 81 [32.1%] vs 36 of 146 [24.7%]), and 33 of 165 adults with dementia (20.0%) compared with 6 of 62 adults (9.7%) without dementia were found to be deprived in 4 dimensions or more. The difference between adults with and adults without dementia in the Multidimensional Poverty Index for deprivation in 4 dimensions was 145.8% for dementia identified by both the AD8 and the RUDAS and 118.2% for dementia identified by either the AD8 or the RUDAS. Education, health, and employment were the main contributors to the adjusted poverty head count ratio. Multidimensional poverty was strongly associated with dementia as measured by the AD8 and the RUDAS (adjusted odds ratio [OR], 2.31; 95% CI, 1.08-4.95), with higher odds for older women (OR, 2.03; 95% CI, 1.00-4.12) or those living in large households (for each additional household member: OR, 1.27; 95% CI, 1.05-1.53). This study suggests that the prevalence and depth of poverty were higher among adults with dementia. A lack of education, poor health, and unemployment were major dimensions of poverty that were associated with a higher prevalence of dementia. Long-term interventions beginning early in life may affect social determinants of health through targeted structural policies (eg, access to quality education and health care) and prevent dementia later in life.