This paper deals with a new phoneme recognition system with information feedback comprising a learning vector quantizer (LVQ) and hidden Markov models (HMMs). This system aims at achieving higher phoneme recognition rates than conventional LVQ-HMM systems, incorporating a feedback of information on the classification of input phoneme which is obtained from the out-put of the LVQ. 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 are higher than those accomplished with conventional LVQ-HMM phoneme recognizers which have no information feedback.