Model‐aided navigation increases navigation accuracy by including a vehicle dynamics model into the filter structure. The commonly used Inertial Navigation System (INS) is hence supplemented by another prediction model for the system state. However, the standard Kalman filter only allows for a single system model to propagate the estimation. The main contribution of this paper is the improvement of an existing approach to estimation with two valid state prediction models. By unifying the models, computation time and state vector size are reduced. Furthermore, the question of how the models must be coupled to achieve optimality and preserve filter stability is addressed.In integrated aircraft navigation, an INS as well as a vehicle dynamics model are available. The presented method unifies these two models and shows superior computational performance compared to existing model‐aided navigation methods and among best results. Furthermore, it is easy to implement and easy to extend with aiding sensors. Copyright © 2013 Institute of Navigation.