Abstract: Classification evaluation of CBM wells provides a basis for the adjustment of coalbed methane reservoirs. However,there exist some disadvantages such as interfering with various factors,and subjectivity of choosing classifica- tion rule during classification evaluation. Besides,it is hard to assess the immediate effect of fracturing on CBM pro- duction,as methane recovery involves a process of drainage,depressurization and desorption. In this paper, an im- proved classification evaluation method of CBM wells was presented,based on data mining and seepage theory. CBM production distribution curve was plot using data through normalization,with normalized well number on horizontal axis and normalized production on vertical axis,whose derivative value was recognized as the criterion of classifying CBM wells. Method of production analysis was then used to evaluate the fracturing effect on different kinds of CBM wells. Classification was carried out in a gas field of Hancheng as well as assessing fracturing effect. Results show that the hy- draulic fracture length of low yield well HS16 was short,skin factor increased during recovery,fracture conductivity was damaged and flow channels was blocked. Possible production after repeated fracturing was 300 m3 / d. High yield well HS14 had a long fracture length and uninfluenced fracture conductivity. Production after fracturing was predicted to be 1 300 m3 / d.