In this paper the direct computing method of precursor information content of anomalies is put forward,and the transformation of non-dimensional precursor information content of precursory anomalies of various observation data has come true. By means of this method we have accomplished the calculation of precursor information content of correlative anomalies which are extracted from the difference values of five-day average from 92 observation items distributed over Gansu,Ningxia and Qinghai.On the basis of reviewed research on the 24 M_s ≤ 5.0 earthquakes we have obtained effective average information contents of unit anomaly about 92 observation items,which become not only important references for earthquake prediction but also a good foundation for the further research on the precursor index system. According to the statistical analysis on the change characteristics of precursor information of M_s ≥ 5.0 earthquakes,the reference indexes of three elements of earthquake prediction are given.