Abstract: Algal blooms is one of the serious environmental problems in our country and even the world at present. In order to reduce and prevent the effects of algal blooms effectively, it collects the meteorological data in meteorological stations of Wuxian, Suzhou from 1986 to 2007, and uses principal component analysis to analyze the main limiting factors and their interrelation one month before the outbreak of algae in Lake Taihu. The average temperature, average pressure, average relative humidity and cumulative precipitation are the main limiting factors in influencing chlorophylla. Combined with the concentration of chlorophyll a from January to October in 2005 in various regions in Lake Taihu, Used the software of Matlab R2010a, the early warning model of algal blooms was established based on BP neural network, this provides scientific basis in order to take appropriate measures to control algal blooms.