This thesis focuses on the development of a health monitoring system for gearbox transmissions. This was accomplished by developing and understanding a two-stage gearbox computer model that emulates an actual gearbox test rig. The computer model contains actual gearbox geometry, flexible shafts, bearings, gear contact forces, input motor torque, output brake torque, and realistic gearbox imbalance. The gear contact force of each gear stage and the input bearing translational acceleration were the main outputs compared between a healthy gearbox and damaged gearbox computer model. The damage of focus was a fatigue crack on the input pinion gear. A sideband energy ratio comparison yielded the computer simulation accurately modeled the difference between a healthy and damaged gearbox. The next step in this study involved the development of a repeatable procedure to initiate and propagate a fatigue crack at the tooth root in an actual spur gear. A damaged spur gear allows for a future comparison of an actual healthy and damaged gearbox system in the lab. A custom fatigue fixture was designed and manufactured for a Martin S1224BS 1 spur gear. The fatigue crack was initiated by position control fatigue testing which deflects the gear tooth a set amplitude for a number of cycles. Over the length of the test, the load that the tooth can withstand in bending decreases as damage begins to occur. Once the max load on the gear has dropped by a significant percentage (5-15%) a crack has initiated and begun to propagate across the tooth face. The use of a scanning electron microscope confirmed the presence a fatigue crack.