The most famous error-decoding algorithm for con-volutional codes is the Viterbi algorithm. In this paper, we present a new reduced complexity version of this algorithm which can be applied to a class of binary convolutional codes with optimum column distances called k-partial simplex convolutional codes.