In this paper, Two-Dimensional Cepstrum (TDC) analysis and its application to word recognition are described. The TDC can represent two different kinds of information contained in speech wave forms simultaneously: static and dynamic information, global and fine frequency structure. Noise reduction filtering or speech enhancement filtering is easily established on this TDC. It is shown that the TDC is an effective parameter for word recognition by both DP-matching and linear matching. Through the word recognition experiments, it is confirmed that the global static information and slow dynamic information are effective for that recognition.