Abstract-A novel multiport matching method is devised to directly maximize the mean capacity with rigorous consideration of the mutual coupling effects of the matching network. In the RF front end of the real communication circuits, the mutual couplings always exist. In this paper, 1) a theoretical capacity upper bound of the 2-by-2 MIMO system with a matching network using the water-filling as the power allocation rule is analytically derived for the first time, 2) the Genetic Algorithm is employed to optimize the parameters of the matching network for the maximization of the mean capacity, 3) a coupled microstrip lines structure is devised to implement the matching network of the real MIMO receiving circuits by this matching method. The numerical results in the last section demonstrate that an optimized matching network obtained using our novel MPM method is capable to enhance the performance of the MIMO systems in a range of different indoor environments. This verifies that our method is not only effective but also practical.