Background: The large-scale clinical problem caused by unacceptable tribological behaviour of certain large diameter metal-on-metal prosthetic hips has directed attention to adverse condition testing. High metal-on-metal wear is connected with adverse reaction to metal debris. Friction is important because high friction may be associated with high wear, risk the fixation of the cup, and cause detrimental heating of periprosthetic tissues. Methods: A friction measurement system was added to a multidirectional, established hip joint wear simulator, and its functionality was evaluated. In preliminary tests, a 50 mm diameter metal-on-metal prosthesis was tested in an optimal acetabular cup inclination angle (48°) and in a steep angle (70°) using a normal peak load (2 kN) and an increased peak load (3 kN). The test length was 100 h. Long-term adverse condition tests of 3 million cycles were run for three 52 mm metal-on-metal prostheses. The lubricant was diluted calf serum at 37 °C. Findings: In the 100 h tests, metal-on-metal frictional torque was not highly sensitive to the angle, load and serum degradation, and it was close to that of a conventional 28 mm prosthesis with a polyethylene cup, mostly below 5 Nm. However, a manyfold higher frictional torque (10 to 20 Nm) was observed in long-term metal-on-metal tests with substantial wear. Interpretation: To obtain a realistic prediction of the frictional behaviour of a hip design, long-term, multidirectional wear tests are necessary. The friction should preferably be measured during the wear test. In addition to normal conditions, adverse condition testing is strongly recommended.