Ball bearings have been used to support rotating shafts in machines such as wind turbines, aircraft engines, and desktop computer fans. There has been extensive research in the areas of condition monitoring, diagnostics, and prognostics of ball bearings. As the identification of ball bearing defects by inspection interrupts the operation of rotating machines and can be costly, the assessment of the health of ball bearings relies on the use of condition monitoring techniques. Fault detection and life prediction methods have been developed to improve condition-based maintenance and product qualification. However, intermittent and catastrophic system failures due to bearing problems still occur resulting in loss of life and increase of maintenance and warranty costs. Inaccurate life prediction of ball bearings is of concern to industry. This research focuses on prognostics of ball bearings based on vibration and acoustic emission analysis to provide early warning of failure and predict life in advance. The failure mechanisms of ball bearings in cooling fans are identified and failure precursors associated with the defects are determined. A prognostic method based on Bayesian Monte Carlo method and sequential probability ratio test is developed to predict time-to-failure of ball bearings in advance. A benchmark study is presented to demonstrate the application of the developed prognostic method to desktop computer fans. The prognostic method developed in this research can be extended as a general method to predict life of a component or system.