An automatic method is developed which extracts of important words from title of paper on the Japanese Classical Literature. The system's object is to extract important words and to accumulate them. For practical use, the extraction algorithm is fixed and tried to shorter run time. In this paper, the fundamental algorithm of extraction of important words is described, and the results is reported which the experiment is done on 2227 titles of paper through a year. The results are fairly satisfactory, and the system is practicable.