The research effort described in this paper is centered on the development and quantitative assessment of a multidisciplinary design optimization environment for the early design phases of expendable launch vehicles. The focus of the research is on the engineering modeling aspects, with the goal of evaluating in detail the accuracy of engineering-level methods for launch vehicle design, both in terms of disciplinary errors (e.g., engine specific impulse evaluation) and of system-level sensitivities, to assess their applicability to industrial early design. Although aerospace applications of multidisciplinary design optimization can be found in literature, the systematic assessment of the models’ accuracies to the extent described in the present research is a rather new endeavor, which is critical for the advancement of this field. In fact, the widespread industrial application of multidisciplinary design optimization has often been obstructed by the difficulty of finding a suitable compromise between the analysis fidelity and computational cost. Considerable effort was therefore spent on a careful, incremental modeling process, with the purpose of overcoming such an obstacle. As a result, although it is clear that the development and tuning of a reliable multidisciplinary environment is a particularly complex and challenging task, detailed investigations showed that a good compromise can indeed be achieved for the expendable launch vehicles application. In particular, the accuracy on the payload performance was assessed to be in the order of 12% for a computational time per design cycle, allowing one to obtain physically sound design changes through the multidisciplinary design optimization, even exploiting only fast engineering-level methods.