The benefits, in terms of social surplus, from introducing congestion pricing schemes in urban networks are depending on the design of the pricing scheme. The major part of the literature on optimal design of congestion pricing schemes is based on static traffic assignment, which is known for its deficiency in correctly predict travels in networks with severe congestion. Dynamic traffic assignment can better predict travel times in a road network, but are more computational expensive. Thus, previously developed methods for the static case cannot be applied straightforward. Surrogate-based optimization is commonly used for optimization problems with expensive-to-evaluate objective functions. In this paper we evaluate the performance of a surrogate-based optimization method, when the number of pricing schemes which we can afford to evaluate (due to the computational time) is limited to between 20 and 40. A static traffic assignment model of Stockholm is used for evaluating a large number of different configurations of the surrogatebased optimization method. Final evaluation is done with the dynamic traffic assignment tool VisumDUE, coupled with the demand model Regent, for a Stockholm network including 1 240 demand zones and 17 000 links. Our results show that the surrogate-based optimization method can indeed be used for designing a congestion pricing scheme which return a high social surplus.