Abstract: The purpose of this study is to describe community structure of heterotrophic bactera in the Donghu Lake (Wuhan, Hubei, P. R. China) and compare empirical taxonomy and cluster analysis methods. A total of 169 bacterial strains growing on nutrient agar plates was isolated from water samples at different stations and water depths. Each isolate was identified by empirical taxonomy. The results indicated that they belong to Achromobacter, Acidaminococcus, Acinetobacter, Aerocoecus, Aeromonas, Alcaligenes, Bacillus, Beneckea, Brevibacterium, Citrobacter, Clostridium, Corynebacterium, Enterobacter, Erwinia, Escherichia, Flavobacterium, Gaffkya, Micrococcus, Microbacterium, Neisseria, Pediococcus, Planococcus, Pseudomonas, Sarcina, Serrtia, Siderocopsa, Siderococcus, Staphylococcus, Streptococcus, and Zymobacterium 30 genera respectively, except for 2 cocci unidentified. The Bacillus was dominanting by number, especially, in the water sample from Station 1, it overran 50% in all strains. It showed that some bacterial groups distribution was different, but the strains of Bacillus, Micrococcus, Acinetobacter, Brevibacterium, Enterobacter, Flavobacterium, and Staphylococcus could be found in all stations. The diversity index of bacterial genus at three stations was approximate to each other. (See tab.4). A cluster analysis by computer IBM PC/XT was used for studying heterotrophic bacterial community structure of the Donghu Lake. Each isolate was examined by conventional techniques for 113 morphological, cultural, biochem-physiological and antibiotic resistance characteristics and by a set of 43 biochem-physiological test system. All data were clustered by computer according to the similarity value. The results indicated the diversity index at Station 1 was higher than those at the Station Ⅱ and Ⅲ, although the diversity index value from the 43 system was lower than 113 system (see tab.4). Comparison of water quality including the concentration of NO2, NO3, NH3, inorganic N, SiO2, total P and N/P showed that the water quality at Station Ⅱ and Ⅲ was similar but was quite different from the Station Ⅰ. (Tab. 3). These data were consistent with the cluster analysis results. In other words, the diversity index from cluster analysis coincided with environmental conditions. Based on above mentioned, it seems that cluster analysis is easier to do than conventional classification. Becouse identification of bacteria is extremely laborious and uncertain due to the geart variability of bacteria, so that that is very difficult in most cases. The chief advantage of cluster analysis is an objective description of individuals, we can get much more information on each strain and consider them as a set of abilities, which can be concerned with thier environment, thus the method may be especially useful in ecological studies.