This study purports to investigate the difference between upward and downward feedback and to find out the strategy to enhance the validity and reliability of the multi-source feedback system. Data were collected from 88 middle-level managers working at divisions of nursing, administration, and clinical technology at a university hospital located in north-eastern part of South Korea. Data about multi-source feedback of the year 2008 and 2009 were obtained from the department of personnel management under the permission of hospital top management. The collected data were analyzed using t-test, and correlation analysis. The major findings of the study were as follows: First, the scores of downward feedback were always higher than those of the upward feedback. Second, the scores of downward feedback were always higher than those of the upward feedback in both nursing and administrative staff, while no significant difference was found in technological staff. Third, the upward and downward feedback scores were significantly different in terms of the evaluation factors. Fourth, there were significant correlations between total scores of downward and upward feedback, and evaluation factor scores. In conclusion, the composition and weight of measurement items of multi-source feedback needs to be designed differently in terms of each job characteristics.