To avoid checking unnecessary or irrelevant conditions of rules, the irrelevant values problem of the decision tree is addressed. We propose an algorithm to remove irrelevant conditions of rules in the process of converting the decision tree to rules according to the semantics of the decision tree. Since our algorithm depends only on the semantics of the decision tree, our algorithm can be integrated into any existing tree-construction algorithm with negligible increase in computational cost concerning that of constructing the decision tree. Moreover, as a side effect, the resultant rules are less likely to suffer from missing branches problem.