Background <p>Conventional musculoskeletal spine models often rely on deterministic kinematic constraints to estimate spinopelvic kinematics and kinetics, which can produce unrealistic intervertebral motions and loading patterns. Incorporating intervertebral stiffness in kinematic evaluations would improve physiological accuracy, particularly during lifting tasks where spinal loading is critical.</p> Methods <p>We developed a stiffness-dependent kinematics estimation workflow that integrates nonlinear intervertebral stiffness within an optimal control formulation, enabling spinal motion to emerge from mechanical behaviour rather than prescribed kinematic constraints in full body musculoskeletal models. Eleven healthy participants performed trunk flexion, lateral bending, and axial rotation tasks, with and without lifting a load. Predicted kinematics and compressive forces were compared against a conventional constraints-driven approach. A single-subject dataset with biplanar radiography provided ground-truth validation of vertebral motions.</p> Results <p>The stiffness-dependent method generated smoother, more physiologically plausible motion distributions and consistently reduced lumbar compressive loading. Vertebral orientation and position errors were lower than with the constraints-driven approach, and compressive forces aligned with literature ranges.</p> Conclusions <p>Nonlinear stiffness modelling yields more evenly distributed spinal kinematics and reduced lumbar loading, providing a physiologically grounded framework for spine biomechanics and injury prevention research.</p>