This study was to identify the levels of perceived stress, immunity cells, physical health and depression, and their relationships among those variables in the elderly who institutionalized comparing home residents. The result of this study can be used as basic data when applying nursing interventions to increase quality of life in the elderly. The questionnaires to estimate stress, health status and depression were collected through direct interview from July to August in 1999 and immunity cells were measured by venous blood specimen collected from 9 to 10 A.M. during the same period. The collected data were analysed using SAS program. The results were as follows. The score of perceived stress of all subjects was 38.49 and perceived stress score of institutionalized elderly(42.62) was significantly higher than that in home resident elderly(34.52). All immune cells tested in this study such as total T cell, helper T cell, suppressor T cell, T4/T8 ration, total B cell, and NK cell, were all under the standard criteria of cells distributions. Most elderly who institutionalized and reside home replied that their health status was not good. However their physical health activity was mostly good even though institutionalized elderly had more disability than home residents. The highest rate was 67.3% as disability due to arthritis. The score of depression in all subjects was 8.2 that indicated having depressive symptom. There was no difference in the depression level between institutionalized elderly and home resided elderly. There was a significant correlationship between physical health and depression, however, the rest of varibles did not show any significant relationships. In summary, the immune cells in the elderly who replied perceiving low level stress, was under normal range. Their health status was perceived as 'not good' but physical health activity was perceived as 'good'. The relationships of stress, immunity, physical health and depression were partially significant but not had evidence as enough as theoretically the suggested relationship. We suggest that further studies using large sample size and more diverse variables should be performed.