Non-fatal injury within industry could be reduced using predictive computer-aided design (CAD) human models to evaluate machinery designs at the concept stage. This paper introduces a constraint-based manikin model that can effectively predict movement patterns over a range of potential designs. The approach was centred upon a proposed methodology that sought key movement events and a movement descriptor as independent inputs to the model. Validation of a generated sit-to-stand case study movement showed the lumbar, hip, knee, and ankle joints to be accurate to within 4.2 degrees across changes in the pace of movement and the designed environment. It is proposed that predictive models can now be generated for other chosen movements by employing the new methodology. Creating realistic CAD-based movement models that change appropriately as the designed environment changes allows industrial workstations and machinery controls to benefit from improved design. Thus, the abilities of the operator can be better matched with the tasks at hand, potentially reducing operator injury when in use.