Abstract: The punctured CCSDS convolutional codes suffered a bit-error-rate performance degradation using the Viterbi decoding algorithm. Aiming at this issue, this paper proposed a max a-posteriori probability decoding algorithm for these codes, it takes a forward and backward update progress of the likelihood messages based on the trellis graph, to obtain the maximum a-posteriori log-likelihood ratio for the corresponding input bits, thus to improve the performance of the punctured convolutional code. As showed by the simulation results, the punctured CCSDS convolutional codes could get an even lower bit-error-rate by using the proposed algorithm, and the higher the code rate, the more significant the bit error rate reduction. Compared with the Viterbi decoding algorithm, the proposed decoding algorithm has a coding gain about 0.2dB and 0.6dB for code rate 5/6 and 7/8 respectively.