It is well-known that the computational complexity of the Viterbi decoding algorithm for an (n,k,ν) convolutional code grows exponentially with k and the code memory ν, and thus it becomes quickly impractical as the code rate increases. A solution, so far, has been to use punctured convolutional codes, which strongly reduces the decoding complexity but leads to a slightly worse performance. Recently, it has been pointed out that every non-punctured high-rate convolutional code (with 2k > 2ν) can be viewed as the concatenation of a block code and a simpler convolutional code. In this paper, we exploit this property of high-rate codes to facilitate the implementation of the Viterbi algorithm. We propose a modification of the Viterbi decoding algorithm for non-punctured high-rate convolutional codes which results in very significant computational savings. Copyright © 2005 AEIT.