This paper deals with a new phoneme recognition system comprising a set of a simplified version of the sub-neural network and the hidden Markov models (HMMs). The ability of this system has been investigated by a phoneme recognition experiment using a number of Japanese words uttered by a native male speaker in a quiet environment. The result of the experiment shows that recognition rates achieved with this system is higher than those attained with the LVQ-HMM system which is known to be one of the most reliable phoneme recognizers.