households using the 2007 Korean Retirement and Income Study. Elderly households were classified into three groups based on the comparisons among the costs of living and then the factors influencing the probability of belonging to each group were investigated using multiple logistic regression models. The major findings are as follows. Firstly, the subjective adequate cost of living was the highest and the subjective minimum cost of living was the lowest. The actual cost of living was in between. Secondly, 42.6% of elderly households belonged to Group1(whose actual cost of living was less than the subjective minimum cost of living), 30.0% was classified into Group2(whose actual cost of living was greater than the subjective minimum cost of living but less than the subjective adequate cost of living) and the actual cost of living of the remaining 27.4% was greater than the subjective adequate cost of living(Group3). Thirdly, income was the strongest factor influencing the probability of belonging to each group, but the influencing factors were different for the logistic models for Group1 and Group3 based on Group2.