Retaining employees in a knowledge-based organisation, such as a university, is a significant challenge, especially as the need to keep knowledgeable workers is key to sustaining their competitive advantage. Knowledge is the organisations’ and employees' most valuable and productive asset, but this intrinsic character leads to a high employee turnover. Often, universities learn about employees' imminent departure too late. To prevent the loss of high-performing employees and to detect the warning signs early, business firms have been using advanced data mining techniques to predict “customer churn”. Recently these techniques have been used with “employee churn” in various industries, but not in higher education. This research bridges this gap by applying data mining techniques to predict employee churns in a university. The contributions of this research will be: 1) to identify critical factors that lead to talent losses; 2) to help universities devise appropriate strategies to retain their employees’ talents.