Active human body models are an important tool to study occupant interaction with safety systems in evasive manoeuvres such as braking and/or steering. In this study a finite element human body model with and without closed-loop active muscle control in the neck and lower trunk was compared to volunteer occupants in six different load cases with lane change, braking, and combined manoeuvres using standard and prepretensioned seat belts. Seven different muscle controllers, using two different muscle activation strategies based either on head and torso displacements or muscle length, and one with the controller turned off have been compared to volunteer kinematics. Cross-correlation analysis with CORA was used to evaluate the model biofidelity. The results show an improvement in CORA scores when using active muscles, compared to the model with muscle activity turned off, for one load case and similar CORA scores between the models for five load cases. CORA scores ranged from 0.78 to 0.88 for the active models and 0.70 to 0.82 from the model with muscles turned off. The active model gave a kinematic response with good biofidelity in lane change with braking, pure braking, and lane change with pre-pretensioned seat belt, but the biofidelity of the model was rated as fair in lane change with standard seat belt.