<b><i>Introduction:</i></b> The long-term cognitive outcomes after transient global amnesia (TGA) have been contradictory in the literature. Our study aimed to longitudinally investigate the association between TGA and incident dementia using long-term data from a nationwide population-based cohort in South Korea. <b><i>Methods:</i></b> The study population was recruited between 2002 and 2020 using the International Classification of Diseases (Tenth Revision; ICD-10) codes from the Korean National Health Insurance Service database. The cumulative incidence curve was plotted to compare the incidence of dementia between the TGA (ICD-10 code G45.4; <i>n</i> = 10,276) and non-TGA (<i>n</i> = 27,389) groups, determined using 1:3 propensity score matching. Using Cox proportional hazard regression models, we obtained crude and adjusted hazard ratios (aHRs) and 95% confidence intervals (CIs) for the incident dementia in patients with TGA compared to non-TGA controls. To examine independent variables determining dementia in the TGA group, logistic regression analysis was performed, and adjusted odds ratios (aORs) and 95% CIs were calculated. <b><i>Results:</i></b> The TGA group had a significantly higher cumulative incidence of dementia than the non-TGA group (<i>p</i> < 0.001, log-rank test). TGA was significantly associated with incident dementia in the univariate and multivariate Cox models (HR 1.34, 95% CI 1.28–1.39 and aHR 1.40, 95% CI 1.34–1.46, respectively). The adjusted logistic regression for incident dementia in the TGA group showed that age (per 1 year, aOR 1.09, 95% CI 1.09–1.10), female sex (aOR 1.31, 95% CI 1.18–1.45), diabetes (aOR 1.21, 95% CI 1.08–1.35), stroke (aOR 1.30, 95% CI 1.16–1.46), depression (aOR 1.53, 95% CI 1.33–1.76), anxiety (aOR 1.24, 95% CI 1.01–1.39), and rural residence (aOR 1.24, 95% CI 1.10–1.41) were independently associated with incident dementia. <b><i>Conclusion:</i></b> Our results suggest a longitudinal association of TGA with incident dementia.