Rigid-body-musculoskeletal models are becoming commonly used to study human motion, particularly of people with some form of neuromusculoskeletal condition. However, these musculoskeletal models are typically generic, having simplified anatomical and kinematic representations, with coarse-meshed bone geometries, and body-segment inertial parameters (BSIP) obtained from a limited number of adult-male cadavers. Furthermore, these models have hip joints that articulate around an estimated hip-joint centre (HJC), two-dimensional-sagittal-plane tibiofemoral (TFJ) and patellofemoral joints (PFJ) models at the knee, and hinges for the talocrural (TCJ) and subtalar (STJ) joints in the ankle and foot. Each person’s model has bone dimensions, BSIPs and joint kinematics that are linearly scaled from the generic models using scale-factors obtained from three-dimensional (3D) positions of motion capture (3DMOCAP) markers or from magnetic resonance images (MRI). However, bone dimensions might not be properly adjusted with linear scaling, which will affect the position and orientation of joint centres and axis, the location of the segments’ centre of mass and moments of inertia, the position of musculotendon length, origins and insertions as well as bone geometry e.g. femur neck-length. Finally, muscle forces, joint angles, moments and internal joint loads of gait estimated using these scaled-generic models have been shown to be inaccurate, probably due to the aforementioned imprecise bone geometries and simplified joint models.