By utilizing the frequency-response-masking (FRM) technique,a novel approach for the design of efficient sharp finite-impulse-response (FIR) digital filters was presented based on a neural-networks optimal design algorithm.The main idea is to minimize the squared-error function in the frequency-domain to obtain FRM filter coefficients.Algorithmic details for the design of basic and multistage FRM filters were presented,the stability theorem was also proved to illustrate the validity of the proposed algorithm,and the implementation of the approach was described together with some design guidelines.Some simulations were included and the results show that the proposed algorithm can design better FRM filters than conventional methods.