The objective of this experiment was to establish the model of near infrared for determining corn stalk fiber quality rapidly, Which is used for strawing fiber quality evaluation of different heterotic groups of maize. The general combining ability(GCA) of fiber quality traits are analyzed, and the lines with good fiber quality and high positive GCA effect are screened to provid reference for the cultivation of lodging resistant maize hybrids. With crushed the third internode and fourth internode of corn stalk as the material, the mathematical models of cellulose and hemicellulose and lignin were established in taking near infrared diffuse reflectance spectroscopy (NIRS) technology. The fiber quality models were used to calculate the stem fiber quality of 94 inbred lines and 376 hybrids, which were planted at two environments a year. Variance analysis of the overall difference in different heterotic groups. The results show that the mathematical models the correction coefficient of determination, cross validation and external are between 0.801-0.998, the error (RMSEC, RMSECV, RMSEP) are between 0.131-1.454. This correction model has high accuracy, which could predict the fiber quality of maize stalk accurately. Therefore，it could provide a reference for the identification of lodging resistant germplasm resources. The fiber quality of LRC is the lowest in these heterosis groups. The fiber quality of P group and TangSPT group are better than other heterosis groups, and it also have a high forward effect in GCA. There are 9 inbred lines in this study, which are not only good quality fiber but also GCA with high positive effect,and there are 2 lines in the P group,1 line in the TangSPT group,2 lines in the Lancaster group,2 lines in the Reid group,2 lines in the Mix group.