Prediction of human movement, and especially of pathological gait, is nowadays an important and mostly unsolved research challenge. In this work, a recently developed computational framework based on optimal control was adopted and explored to assess its potential for predicting a pathological gait pattern, in particular the crouch gait typical of subjects affected by cerebral palsy. To this end, the generic musculoskeletal model on which this optimal control framework is based was made representative of such pathological case by modeling contracture of relevant muscle groups commonly associated with crouch gait, namely knee and hip flexors. All the conducted simulations succeeded in inducing the model into a crouch gait pattern, despite their diversity in cost functions. Moreover, the obtained joint angle trajectories correlated well with the experimental ones obtained from a CP child walking in crouch. These kinematic results suggest that optimal control techniques and proper tuning of musculotendon parameters are an important pairing for predictive simulations of human walking. On the other hand, the obtained results confirm that estimation of muscle activations is strongly dependent on the selected objective function and still requires deeper investigations.