Blended wing body is an emerging tailless concept for aircraft design. The main challenge in developing this new concept is its flight stability. An efficient approach to resolve this challenge is through numerical simulations. The multidimensionality of the design space results in high computational loads which brings us to build approximations to the general computer input/output relationships known as metamodeling or surrogate modelling. Here polynomial and Gaussian regressions are experienced and assessed for their relative efficiency. It is shown, in the present case that, on one hand, Gaussian interpolation with careful sampling is the most efficient, and on the other hand, polynomials are easier for understanding functional behavior and its evaluation.