Abstract: By means of single item index , 7 seismological indices, including frequency, earthquake creeps, b v alue, averag e mag ni tude, Zvalue, GL value and modulation ratio, are selected for synthetical analysis in order to extract the further precursory informatio n. The time-space scanning is made for all seismological indices. Considering the fuzzy relation between space uni tanomaly and the location of moderatestrong earthquakes, the condition probability for occurrence of moderately st rong earthquakes after the anomaly appearance in each unit has been obtained. Then the index weighting of the synthetical probability is inferred based on the prediction ef ficiency of each sing le index in the history , from which the weighted synthetical probability for occurrence of moderately strong earthquakes is obtained by using the dif ferent space units and different time window according to Bayes rule. The results (since 1970) indicate that R-value of prediction efficiency of synthetical probability method is g reater than 0.5.

Key words: Seismological index, Prediction efficiency, Fuzzy relationship, Condition probability, Weighted synthetical probability