This paper presents a computationally efficient design of modified cosine-based decimation filters. One of the main contributions of this paper is the proposal of a multiplierless finite impulse response low-order linear-phase filter to increase spurious signal rejection in the so-called folding bands. The resulting filters feature reduced computational complexity compared with other recent proposals in the literature, as well as higher folding-band attenuation than traditional comb filters. The frequency behavior of the proposed class of decimation filters is compared with generalized comb filters as well as other proposals in the recent literature. Moreover, this paper addresses field-programmable gate array implementation of the proposed filters in a variety of structures, including both nonrecursive and recursive architectures, and it evaluates and compares the performance of these architectures in terms of area and power consumption. A critical comparison is provided with the goal of highlighting key design issues and computational efficient implementations for the interested reader.