Due to the ever growing demand for more efficient aircraft novel aircraft concepts have to be explored. By improving design tools the potential of unconventional configurations can be further studied. This requires improvement of conceptual design tools such that more knowledge can be gathered on alternative solutions as early in the design process as possible. Multidisciplinary design optimization (MDO) can support this process by providing an environment in which the various disciplines can be designed and optimized concurrently, while a certain level of consistency is maintained. An optimization design tool has been created to assess the potential performance gains of novel aircraft configurations. It connects with the Initiator design tool, which is a conceptual design framework. As such, it can also be used as a means to expose any design issues that may exist in the Initiator. With the optimizer tool the following four case studies were performed: a conventional Airbus A320, a forward-swept canard aircraft, a threesurface aircraft and an oval-fuselage aircraft. For this purpose the genetic algorithm, a gradient algorithm and a hybrid genetic algorithm were used. From the case studies followed that large improvements can be obtained with unconventional aircraft configurations when compared to the initial aircraft design proposed by the Initiator design tool. Up to 20% improvement was found with the three-surface and canard aircraft. The oval-fuselage aircraft could be improved by a solid 10%, while a 5% improvement was obtained with the conventional A320. Among all cases the most contributing factors were the wing position, sweep angle and aspect ratio. There is a tendency towards lower sweep angles due to the positive effect on the weight of the wing and an underestimation of the drag rise. With the forward-swept canard relatively high sweep angles were found, from which followed that the weight penalty of forward swept wings is underestimated. The sizing routine of the control surfaces is found to be inadequate, since the Initiator derives most parameters directly from the wing and does not properly take into account control and stability requirements. Results have shown that this mainly regards the sweep and dihedral angle. These sizing issues also affect the static margin. It was found that class II design information was not fed back to the control surface sizing. From the used optimization algorithms can be concluded that the gradient algorithm was the least effective as it had difficulties with the noise. It sometimes stopped prematurely or started oscillating. The genetic algorithm was found to be the best option due its robustness. It proved to be far less sensitivity to noise. Its computational cost could be significantly reduced by applying parallel optimization and using a caching mechanism. The hybrid algorithm was found to be too computational expensive. The obtained increase in objective value did not outweigh the added cost.