As people get aged, we need to pay more attention to the elderly living condition with respect to welfare policy. The present study focuses on housing condition of elderly people living in rural area, and analyzes its determinants, periodical difference, and regional disparities, incorporating an index of minimum housing standard. This study applies multi-level logit model that has a strong statistical advantage that can take random aspects of household and regional context into account. We found that married couple household, presence of economically-active family members, higher education, specialized profession, migration and higher level of land price tend to lower the probability of residing in substandard housings. We also found that housing conditions for elderly people prove to be different by regions and it is particularly noticeable in mountainous rural areas in Gyeongsangnam-do and Gyeongsangbuk-do with the poorest elderly housing conditions. Results also implies that housing welfare policies should be implemented differently considering the target groups based on regional differences. We conclude with some additional policy implications for the elderly living in rural areas.