This study was to evaluate hospital characteristics as composition of manpower and facilities to the death rate of patient; and to earmark the factors affecting the overall hospital mortality rates. The data utilized were derived from survey material conducted by the Korean Hospital Association on 32 tertiary referral hospitals in Korea between 1986 and 1994. The findings are: 1. Those hospitals having the most capacity per bed had little difference to the mortality rates than the others. 2. Those hospitals having the most daily patients per specialist had significantly higher mortality rates than the others, but the number of daily patients per nurse had little effect on the mortality rates. 3. Those hospitals which had a relatively sufficient number of quality assurance activities revealed a lower mortality, and particularly in case where such effort was directed to the clinicians, the outcome was remarkable. we concluded that the major factor affecting the hospital mortality rates seems to be the number of specialists per number of beds, the degree of quality assurance assessment of the clinicians, the quality assurance activities of each hospital as a whole, and the number of daily patient per specialist. According to the findings of this study, the composition and quality of specialist and adequate quality assurance activities seemed to be the essential for the improvement of hospital care. Therefore, in this regard the proper implementation of policy and support is highly recommended. Due to lack of available research material, the personal characteristics of specialists haven't been considered in this study However, this longitudinal observation of 32 tertiary referral hospitals over a nine year period has significant merit alone.