There are several approaches to reasoning from an inconsistent knowledge base, such as the consistency-based method and the argument-based method, from the point of view of the definition of conclusions derived from an inconsistent knowledge base. This paper proposes a treatment focusing on the condition for assuring the validity of conclusions derived from an inconsistent knowledge base. A consistency-based method performs deductive reasoning with consistent subsets selected from an inconsistent knowledge base. There exist maximal consistent sets which derive different conclusions inconsistent with each other. Thus, we propose a condition to distinguish these sets in order to assure the validity of conclusions. On the other hand, an argument-based method takes an argument that consists of a conclusion and a consistent knowledge base which derives the conclusion. This method selects an acceptable argument according to the alternative relation of possible arguments, as a conclusion has arguments from which it is derived and alternative arguments. Thus, we propose a condition which undercuts alternative arguments, in order to assure the validity of a conclusion. We show that these two methods are essentially identical with regard to the validity of a given conclusion by proving these conditions' equivalence. © 2006 Wiley Periodicals, Inc.