Abstract. In aeronautics, the first design stages usually involve to solve a constrained multi-disciplinary optimization problem. The Bayesian optimization strategy is a way to solve such a complex system. This approach requires to evaluate the objective function and the constraints quite a few times. Evaluations are generally performed using numerical models that can be computationally expensive. To alleviate the overall optimization cost variable information sources can be used to make the evaluations. Typically we are dealing with cheap low fidelity models to explore the design space and expensive high fidelity models for exploitation. In the following work, a mono-fidelity Bayesian optimization method and its multi-fidelity counterpart are compared on two analytical test cases and on an aerostructural drone design constrained optimization problem. The multi-fidelity strategy allows to divide the computational cost by 1.3 compared to the mono-fidelity one on these test cases