The monitoring techniques and methods of Verticillium wilt were benefit for increasing yield and efficiency of cotton, and could provide theoretical basis for distribution of crops and disease resistant variety. The study make use of TMsatellite multispectral image in the study area as data sources, combining with the ground survey data, find the optimal band combination to monitor cotton fields infected Verticillium wilt. Then four supervised classification methods, included minimum distance method, the parallelepiped method, spectral angle mapping classification and support vector machine algorithm, were applied to recognize cotton fields of Verticillium wilt. Results showed that false color band combination which from the blue band (band1), near infrared wave band (band4) and the short infrared wavelengths (band5) of the multispectral image, can be used as optimal combination of TM image to monitoring cotton fields of disease. Cotton fields of diseases could all been recognized and classified into different types by four supervisedclassification methods during blooming period; and the results of the parallelepiped method was most closest to reality, the overall accuracy and kappa coefficient were 90% and 85% ,respectively, were highest in the four algorithms. The results could satisfy the production requirements, and be carried out in fast diagnosis of cotton field infected Verticilliumwilt.