Taking into account the considerable complexity and difficulty of conventionally proposed nonlinear system identification methods and techniques by discrete Volterra functional representation, a simpler practical identification method is proposed. In the proposed method, discrete Volterra kernels, divided into groups corresponding to each sampling time, are regarded as unknown variables of each simultaneous linear equation. These equations have coefficient matrices constructed by combining the sampled discrete input signals, and by computing constant vectors derived from the output excluding the contribution of the input signals which came in, long before the output in the identification procedure. A group of simultaneous linear equations is solved one by one, applying the solution of each step to solve the next simultaneous linear equation having a larger number of unknowns, and thus each identification procedure reduces the size of the matrix to be dealt with. This iteration process assures a highly accurate identification of a system and it can be widely applicable because it allows a choice of input signals and systems to be identified. This can be applied even to systems in operation, if appropriate combinations of input signals are selected, which can be automatically performed by numerical determination under the condition of their linear independence. Furthermore, a more practical method is proposed, in which the number of the test signals can be reduced in case that the appropriate combinations of independent input signals are difficult to obtain, especially in the system in operation. For the explanation of the methods two mathematical systems are identified by the proposed methods and their identified Volterra kernels and input-output relations are shown.