With enhancement of living quality, general public has pursued better patient-centered service when seeing a doctor. Besides professional medical treatments, the overall healthcare quality receives more attention these years. While nursing staff plays a critical role during the whole medical process, its caring attitude, patience, and attentiveness greatly influence the patient's satisfaction, leading to mutual trust between the medical staff and the patients. High turnover rate of the nursing staff, ranging from 15% to 40% each year, however jeopardizes consistency of its services. Many factors can raise such instability, such as long and tiring working hours, potential life-threatening injuries at work place, lack or insufficiency of colleague supports, mental and physical maladjustment towards realistic work environment, especially for the newcomers, etc. Early detection of the potential nursing staff turnover will thus not only reduce the recruiting and training costs, but also help render assistance to the nursing staff and promote the patient's satisfaction as well as the healthcare quality. This study employs a decision tree algorithm, with respect to the basic personnel information, to correlate tendency of nursing staff's departure. A public hospital in northern Taiwan participated as the research subject: 2351 personnel samples were collected as they were in July 2011. With parameter optimization and rule-reasoning association, the proposed decision tree algorithm exhibits an accuracy rate of 78.81% of nursing staff turnover prediction, and offers recommendations of alleviating potential departures. The results of the study can serve as a great reference for the hospital administration and could be applicable to the other medical institutions.