The knowledge acquisition is one of the most difficult and time-consuming activities in developing knowledge-based systems. In this work, we propose a novel knowledge acquisition method through documents analysis. The knowledge base can be built correctly, rapidly, and partially automatically by using this method. This method is especially useful when it is difficult to find domain experts. We apply this method to the knowledge acquisition for a simple dynamic alarm processing system (DAPS) for nuclear power plants and develop the alarm knowledge acquisition system (AKAS) for DAPS by using G2. (C) 1998 Elsevier Science Ltd. All rights reserved.