The relation between the performance of AdaBoost and that of component classifiers is analyzed, and the approach of improving the classification performance of RBFSVM is studied. There is an inconsistency between the accuracy and the diversity of component classifiers, and the inconsistency affects the generalization performance of the algorithm. A new variable ó - AdaBoostSVM is proposed by adjusting the kernel function parameter of the component classifier based on the distribution of training samples, and it improves the classification performance by making a balance between the accuracy and diversity of component classifiers. Experimental results indicate the effectiveness of the proposed algorithm.