The Monte Carlo-based Characteristic Basis Finite-Element Method (MC-CBFEM) is developed for predicting the statistical properties of the 2-D electromagnetic scattering from objects (such as ship-and decoy-like objects) on or above random rough sea surfaces. At each realization of the Monte Carlo technique, the 1-D rough sea surface is randomly generated by using the Pierson-Moskowitz spectrum, and the bistatic radar cross section (RCS) is computed by employing the CBFEM approach. The CBFEM is a noniterative domain decomposition finite-element algorithm, which is designed to alleviate the challenges of the conventional finite-element method in solving large-scale electromagnetic problems. The CBFEM partitions the problem into a number of nonoverlapping subdomains and generates physics-based characteristic basis functions for the representation of the fields in each subdomain. Since this approach reduces the matrix size and lends itself to convenient parallelization, it is attractive for efficiently solving large-scale problems many times in the Monte Carlo simulation with the use of direct solvers and small-sized matrices. For a number of surface realizations, each of which can be considered as a sample from the random process specifying the surface, a family of bistatic RCS values is obtained as a function of incidence angle and surface roughness (or wind speed). The coherent (mean) and incoherent (variance) components of the RCS are illustrated with particular emphasis on the effects of surface roughness and the angles near grazing. Statistical characterization is also achieved by other means, such as correlation coefficient and density functions represented by histograms.