The underlying basic model represents a powertrain with automatic transmission including a torque converter. It is based on a greybox-modeling approach, which refers to ordinary differential equations with identified parameters and characteristic curves. The validated basic model is extended in order to reproduce the system behavior and especially the side shaft torque during a gear shift process. Therefore the model is extended by a transmission model with clutches for gear shifting in order to simulate specific powertrain dynamics additionally. The parameters have already been determined for the basic model using a method for isolated and structured parameter identification based on measurement data of an automotive powertrain test bench. A comparable structured parameter identification method is applied to obtain the parameters of the extended model.The extended model will be compared with a state observer which is based on a low-order model and neglects the detailed consideration of submodels with nonlinear characteristics. The observer also reconstructs the side shaft torque during the entire maneuver and will be consulted as an alternative approach for estimating non-measurable states in the system. Comparing the extended model with the observer model will illustrate the limits and potentials of both approaches. The simulation results are compared and evaluated based on the measured side shaft torque.Within this study the extended powertrain model is used as reference model that describes the vehicle dynamics for the considered maneuver precisely, whereas the observer is an alternative approach trying to reach similar results mainly obtained from the measured speed signals.