Two algorithms for identifying bacteria are investigated. One is the distance analysis with pattern slide method, and the other is the power-spectrum analysis of the differential pattern of the growth curve. Bacterial growth is measured by monitoring the turbidity change of culture medium. An automated system of measuring and data processing is constructed for the identification of bacteria by microcomputer, turbidity detecting apparatus and some electrical circuits. For ten species of coliform bacteria, 93.3% of tested 178 isolates and 95.0% of tested 190 isolates have been correctly identified by the distance analysis and the power spectrum analysis, respectively. It is contended that multi-channelization of the system is essential as it leads to rapid and reliable identification of bacteria.