In this work the inverse problem solution with iterative Gauss-Newton algorithm and Truncated Singular Value Decomposition (TSVD) is shown. For the goal function a norm l2 was chosen. To solve the inverse problem, which consists of the identification of conductivity distribution in a 3D model, the multi-frequency sensitivity analysis has been applied. The correctness of sensitivity calculation has been proved utilizing three different methods, namely Tellegen's method of adjoint model, differentiation of stiffness and mass matrix, as well as sensitivity approximation by means of difference quotient. Regarding the effectiveness of those methods, the first one is preferred because of shortest computational time.