This study is about major symptoms of elderly and medical services for elderly in long-tenn care facilities. The subject of this study was 298 patients over 00 years old staying in two geriatric hospitals and two nursing homes. The symptoms and medical services were level of patient classification from RUG(Resource Utilization Group)-III which is applied for both Medicare and Medicaid for skilled nursing facilities reimbursement system in US and designed for measuring patient characteristics and medical staff time. This classification is explained by each patient resource(staff time) utilization level which is called CMI(Case-Mix Index). In this study, the symptoms and services were compared by facility type and they were categorized by level and compared by CMI. Major findings are as follows; 1. There were more elderly who have cognitive function problems in nursing homes than patients in geriatric hospitals. There were more patients with behavioral problems in geriatric hospitals than residents in nursing homes. These results were both statistically significant. 2. The patients in geriatric hospitals received significantly more nursing rehabilitation services, rehabilitation services and extensive services than residents in nursing homes. Other hands, special care services were provided significantly more to residents in nursing homes than elderly in geriatric hospitals. 3. ADL and depression variables had higher CMI when the symptoms were heavier condition. The CMI were not matched with levels of cognitive function problems and behavioral problems. 4. The CMI matched well significantly with levels of nursing rehabilitation services, special care services, and clinically complex services provided for the patient in geriatric hospitals and only nursing rehabilitation services in nursing homes. The CMI for rehabilitation services level and extensive services had regular trends. From the result of this study, the resource utilization level and services provided for elderly in each long-term care facilities were figured out. For the further study, it needs to have more concern about RUG-ill which classification variables were just analyzed.