The 16-station RandomPOD wear test system, previously validated for prosthetic hip wear, was used in the simulation of knee wear mechanisms with a ball-on-flat test configuration.This consisted of a CoCr pin with a ground and polished spherical bearing surface (radius 28 mm) against a conventional, gamma-sterilized UHMWPE disk in serum lubrication. The biaxial motion, consisting of x and y translations, and the load were non-cyclic. Relative to the disk, the center of contact wandered within a circle of 10 mm diameter, and the average sliding velocity was 15.5 mm/s (range 0 to 31 mm/s). The load varied non-cyclically between 0 and 142 N (average 73 N). In the 60-day test with 16 similar wear couples, moderate adhesive wear, the principal wear mechanism of a well-functioning prosthetic knee, dominated. This showed as a burnished, circular wear mark (diameter 13.2 mm, area 137 mm 2 ). The wear factor was 2.04 ± 0.03 × 10 -6 mm 3 /Nm (mean ± 95 per cent confidence limit). For the first time a truly multidirectional, realistic and uniform, large capacity pin-ondisk simulation of knee wear mechanisms was implemented.