Interaction of nominally flat engineering surfaces that leads to a large contact area exists in many mechanical systems. Considering periodic similarity of surface geometry, a numerical three-dimensional elastoplastic contact model can be used to simulate the contact behaviors of two nominally flat surfaces with the assistance of the continuous convolution and Fourier transform (CC-FT) algorithm. This model utilizes the analytical frequency response functions (FRF) of elastic/plastic responses of materials and provides contact performance results, including the average surface gap, the contact area ratio, and the volume of plastically deformed material, which may be defined as performance variables. Following the digital filtration technology, rough surfaces can be numerically generated with specified autocorrelation length and the first four orders of statistical moments. A group of contact simulations are conducted with various working conditions. The effects of topographic and material properties on the contact behaviors are discussed. With a multivariables regression method, empirical formulas are developed for the performance variables as functions of surface statistical characteristics, material properties, a hardening parameter, and the applied load in terms of pressure.