Intermediate frequency (IF) filters find diverse practical applications in modern communication systems. This paper presents a novel technique for the design and optimization of digital IF filters. This technique consists of two separate stages. In the first stage, the bilinear-LDI lattice digital filter realization approach is exploited and applied to the design of an infinite- precision digital IF filter responsible for satisfying the given design specifications only approximately. In the second stage, a diversity-controlled (DC) genetic algorithm (GA) is applied to the finite-precision optimization of the resulting IF digital filter so as to satisfy all the design specifications exactly. The DCGA optimization employs an existing cross-generational probabilistic survival selection scheme for increasing the convergence speed. The application of the proposed technique to the optimization of an eighteenth-order AMPS digital IF filter operating around a center frequency of 455 kHz only required 500 generations to converge.