Abstract: Impervious surface is a significant indictor in monitoring eco-environmental health.In linear spectral mixture analysis(LSMA) model,four endmembers are selected.Low albedo and high albedo,the factor of impervious surface are hard to obtain precisely.In order to overcome the flaw of LSMA,fuzzy ARTMAP was proposed to estimate the percent of impervious surface(ISP).The method included the multi-resource information from Landsat ETM+ and Quickbird and the superiority of auto-adapted of Fuzzy ARTMAP neural network.Taking the city of Wuhan in Hubei province as the example,the result showed that the precision of ISP estimated by fuzzy ARTMAP was higher than that of LSMA.The distribution of impervious surface obtained by fuzzy ARTMAP was consistent with the actual land surface.