To evaluate how much power a non-metallic gear (the term non-metallic gear is used to cover both polymer and composite gears in this thesis) can transmit, it is necessary to investigate both the thermal fatigue strength and wear resistance of the gear. This thesis presents an investigation into polymer and composite gear strength and wear together, focusing particularly on the thermal behaviour of acetal gears. A literature survey is presented, which includes the design, performance and thermal behaviour of non-metallic gears. A novel test rig has been successfully designed and developed for the measurement of the wear and life of non-metallic gears. The loading method permits a large amount of tooth wear without significantly affecting the applied load- a feature unique to this configuration of test rig. A computer based, real-time measurement system has also been developed which allows the continuous measurement of gear wear and which records the time to failure. This continuous measurement of non-metallic gear wear is also unique and reveals aspects of nonmetallic gear behaviour not reported elsewhere in the literature. Wear and life test results at four running speeds, 500,1000,1500 and 2500 revs/ min, for acetal (Delrin 100), nylon 66 and filled composite (55% nylon, 30% glass fibre, 15% PTFE) gear pairs are presented. These results gave some important clues to the ways in which non-metallic gears fail, wear, scuffing, fracture at the tooth root and fracture at the pitch point are shown to be the main causes of failure. The test results are examined and compared with existing non-metallic gear design standards. The wear rates measured by previous researchersu sing acetalr unning thrust bearing wear testsa re also comparedw ith acetal gear wear results. The behaviour of acetal gears in particular was investigated further. As a result of this a critical torque referred to as the transition torque, has been identified for acetal gears beyond which the wear rate rapidly accelerates, causing the scuffing on the gear surface. A step loading method has also been developed by which the transition torque can be quickly measured and with reasonable accuracy. Tests have proved that acetal gear scuffing is caused by the maximum gear tooth surface temperature reaching the melting temperature of acetal material. To predict the transition torque it is essential to find the maximum gear tooth surface temperature. This temperature can be divided into three components; ambient, bulk and flash. These three components were investigated in detail. The assessmenot f the effects of ambient temperatureo n acetal gear performance enabled the variation of transition torque with ambient temperature to be studied. Gear body temperatures were measured using an infra-red thermal image camera. To find the convection heat transfer mechanism, the air temperature around the gears were studied. As a result of this a simple heat dissipation model is suggested enabling the bulk temperature to be predicted. The results are in close agreement with the measured data obtained using the thermal image camera. A theoretical analysis of the flash temperature and the partition of heat between the driver and driven teeth has also been made. A new design method for acetal gears is suggested in which the gear surface temperature is first predicted and then compared with the melting temperature of the material. The transition torque determines the load at which the surface temperature reaches the material melting temperature and provides scuffing limits for the gears. This method has been proved by tests on gears of modules I and 2 mm and at different running speeds. Finally, general conclusions are given and some recommendations for future work are suggested