Objective The aim of this study was to assess the accuracy of an imageless navigation system in localising the hip centre and to evaluate the effect of pelvic movement on the accuracy of hip centre acquisition. Material and methods A phantom leg was constructed to simulate the hip joint and upper femur. A 3D measurement device was used to measure precisely the co-ordinates of hip centre. A limber link simulated hip motion as used when defining the hip centre during computer-assisted surgery. The data generated by the 3D measurement device and the image-free navigation system was compared with increasing amounts of simulated pelvic movement. Results For moderate pelvic movement (5 mm) the mean error of the hip centre ranged between 1.5 mm and 3.9 mm with a corresponding angular error of between 0.25° and 0.64°. In conditions of excessive pelvic movement (15 mm) the mean error was 11.7 mm corresponding to an angular error of 1.9°.