Naive Bayes classification is a kind of simple and effective classification model. However, the performance of this model may be poor due to the assumption on the condition independence. By introducing association rules, this classification model can be improved in two way. On the one hand, the associated relationship between condition attributes can be found out through association rules mining, in order to weaken the independent assumption. On the other hand, Naive Bayes is weighted by computing the confidence of association rules.