This paper proposes the equivalent linearization (EL) and sliding mode control (SMC) methods to address nonlinearity and enhance the performance of nonlinear systems subjected to nonstationary random excitations. The EL methods are commonly used to propose approximate solutions for nonlinear systems under random excitations due to the high computational demands in the exact analyses. Considering the application of orthogonal functions in improving the accuracy and efficiency of approximate solutions, an orthogonal block pulse (BP) function is applied to the EL method in this study to approximate the nonlinear system responses under nonstationary random excitations. The SMC, as a robust control method for systems with uncertainties and external disturbances, is capable of achieving reliable and accurate tracking control. This method is applied to effectively reduce the dynamic responses predicted by the proposed EL method under nonstationary random excitations. The proposed approach is tested on single-degreeof-freedom and two-degree-of-freedom Duffing systems subjected to a seismic type excitation. The results indicate that not only the orthogonal functionbased EL method can approximate the dynamic responses more accurately, at lower computational cost, and by a high agreement with the exact solution, but also the proposed SMC can improve the performance of nonlinear systems by effectively reducing the responses compared with the linear quadratic regulator control method.KEYWORDS block pulse functions, equivalent linearization, nonlinearity, sliding mode control Nomenclature: A, plant matrix of the system; B,B u , control location matrices; B r , excitation orientation vector; C,C eq , damping matrix and its equivalent value; E[·], mathematical expectation; g, nonlinear function of displacement and velocity; h, unit impulse-response function; J, cost function; J G ,J F , convolution operational matrix; K,K eq , stiffness matrix and its equivalent value; k, linear stiffness; M, mass matrix; m, positive integer value; P, sliding surface coefficient vector; Q,R, positive semidefinite and definite weighting matrices; q, width of block pulse