Abstract: Electromagnetic Nondestructive Evaluation (ENDE) is applied in various industrial domains for the exploration of hidden in-material defects of structural components. The principal task of ENDE can generally be formalized as follows: an unknown defect affects a given host structure, interacting with a known electromagnetic field, and the response (derived from the electromagnetic field distorted by the defect) is measured using one or more receivers at known positions. This response contains some information on the electromagnetic constitutive parameters and the geometry of the defect to be retrieved. ENDE aims at extracting this information for the characterization of the defect, i.e., at the solution of the arising “inverse problem”. To this end, one has to be able to determine the electromagnetic field distorted by a defect with known parameters affecting a given host structure, i.e., to solve the “forward problem”. Practically, this is performed via the mathematical modeling (based on the Maxwell's equations) and the numerical simulation of the studied ENDE configuration. Such simulators can provide fine precision, but at a price of computational cost. However, the solution of an inverse problem often requires several runs of these “expensive-to-evaluate” simulators, making the inversion procedure firmly demanding in terms of runtime and computational resources. To overcome this challenge, “surrogate modeling” offers an interesting alternative solution. A surrogate model imitates the true model, but as a rule, it is much less complex than the latter. A way to construct such surrogates is to perform a couple of simulations and then to approximate the model based on the obtained data. The choice of the “prototype” simulations is usually controlled by a sophisticated strategy, drawn from the tools of “design-of-experiments”. The goal of the research work presented in this Dissertation is the improvement of ENDE methods by using surrogate modeling and design-of-experiments techniques. Three self-sufficient approaches are discussed in detail: an inversion algorithm based on the optimization of an objective function and two methods for the generation of generic surrogate models, both involving a sequential sampling strategy. All approaches presented in this Dissertation are illustrated by examples drawn from eddy-current nondestructive testing.