Injection-moulded composite gears using fibre-reinforced thermoplastics often exhibit increased performance compared with those made from unreinforced polymers. In general, materials are chosen on cost or simple performance grounds (such as, nominal flexural modulus or heat deflection temperature). Aspects of the performance of polymeric gears have been studied by a number of workers and efforts have been made to simulate the contact conditions during gear running. However, until now the wear performance of gears made from polymer matrix composites has not been studied systematically. While such materials have been studied using pin-on-disc or twin disc roll/slide wear techniques, no attempt has ever been made to directly compare the results from such studies with those from gear tests. This paper attempts to explain the comparative methods of measurement of various polymer matrix composite gear materials and to relate their performance to results obtained in contact simulation experiments by other workers. Methods of wear testing are compared including direct gear testing and disc testing, together with electronic (displacement) measurement, weight loss and direct measurement. A new method of characterising the wear of gears is presented, which relates actual contact conditions and gear tooth wear. This co-ordinate measurement technique provides wear rates as a function of roll angle, and hence load, sliding speed and slip ratio. It allows comparisons between gear tests and a simpler test configuration, the twin disc roll/slide test. The conclusions reached have interesting commercial implications for the producers and end users of polymer composite gears and for those involved in testing them.