A maximally flat FIR filter design method based on explicit formulas combined with simulated annealing and random search was presented. Utilizing the explicit formulas to calculate the initial values, the finite-word-length FIR filter design problem was converted into optimization of the filter coefficients. An optimization method combined with local discrete random search and simulated annealing was proposed, with the result of optimum solution in the sense of Chebyshev approximation. The proposed method can simplify the design process of FIR filter and reduce the calculation burden. The simulation result indicates that the proposed method is superior to the traditional round off method and can reduce the value of the objective function to 41%-74%.