This paper proposes the use of gray box mathematical modeling, through Identification Systems theory, for obtaining the behavioral performance of micro electro mechanical systems (MEMS) device. The target structures are two elastic masses, called double bridge and hinge, installed in a comb-drive actuator with simple typology and unidirectional displacement. The model structure used is the autoregressive with exogenous inputs (ARX), in which the parameters are estimated using the least squares method. The ARX estimated models, for both elastic masses, are compared to the MEMS structures simulated in the ANSYS computational tool, considering the relative error percentage and cross-validation indicators, and the root mean square error, Akaike information criterion, and Bayesian information criterion indexes. From the simulation results, it was observed that the ARX estimated models, for both elastic masses, presented a satisfactory accuracy. Therefore, the use of mathematical modeling can be suggested for MEMS device designers in order to contribute to the behavioral performance and the quality of these microstructures, reducing the production costs and manufacturing time.