Reconstruction of the human body movements is a challenging research field with several application scenarios. Despite some of them are characterized by low-fidelity constraints, some others require high accuracy and precision. This is the case of quantitative evaluation of joint function, rehabilitative interventions and other biomedical applications. Joint function is often modelled as a spherical joint model. The position of the corresponding center of rotation can be obtained through ad-hoc anatomical calibration based on the recorded joint motion (functional approach). In this paper, we investigate the effects of some protocol parameters (speed, movement type and range of motion) on the estimation of the joint rotation center when using magneto-inertial sensors. The evaluation has been carried out on a mechanical analogue of the gleno-humeral joint. The results revealed that the most critical factor is the speed of the joint movement execution. Movement type and range of motion have a minor impact. The best CoR position estimates were obtained for the fast circumduction motions (4±1 mm). The order of magnitude of the error achievable with this method is the same of that found with stereo-photogrammetric technology, with the advantage of wearability and occlusion-free recordings.