A new algorithm for detecting and/or decoding of signals in communication system is presented. The advantage of the proposed suboptimum algorithm with respect to the classical maximum likelihood detection (MLSE) method lies in the significant reduction in computation complexity attainable by associating a reliable information to each received signal. At the same time, the error probability is maintained close to MLSE levels through the proper selection of the reliability information. We describe the application of the proposed algorithm to some interesting cases, such as the detection of continuous phase modulations with and without combined channel coding, detection of Ungerboeck codes, use in conjunction with the Viterbi algorithm. In all cases, a net reduction in computation complexity with respect to the classical MLSE algorithm is achieved.