Objectives : To identify the risk factors of dementia among the elderly in a large city. Methods : A cross-sectional study was conducted in July 2001, with potential participants selected by stratified two stage cluster sampling of the elderly population of Keumgog dong, Busan. A total of 452 elderly people aged 65 years and over, underwent a two phase diagnostic procedure. Mini-mental State Examination-Korean (MMSE-K) and Samsung Dementia Questionnaire were used for the 1st stage, and the Clinical Dementia Rating Scale (CDR), the Bartel ADL, and IADL Index, the Korean Geriatric Depression Scale (KGDS), the Modified Hatchinski Ischemic Scale (MHIS), and other laboratory tests were used for the 2nd stage. Results : Of the 446 participants finally chosen, 45 were confirmed with dementia, and 363 as normal, with the rests not confirmed with dementia or as normal, were excluded from the analysis. According to the logistic regression analysis, the risk of dementia was significantly higher In: people aged 80 and above (OR=4.36, 95% CI=1.97-9.62), illiterate (OR=3.58, 95% CI=1.71-7.46), who had a history of strokes (OR=6.35, 95% CI=2.71-14.87), or who had 3 history of hyperlipidemia (OR=4.74, 95% CI=1.65-13.61), compared to their counterparts. Conclusions : These results suggest that efforts to prevent strokes and hyperlipidemia can significantly decrease the risk of dementia.