Abstract: The intact rock strength, RQD, jointing space, friction angle of discontinuity and so on are combined to evaluate the cavability of an ore body to be mined by block caving. The data acquisition and processing system of these indexes are discussed. Applying MicroMine software, a 3D block model of the whole ore body is established. Based on the value of each parameter in a block estimated by geostatistical technique, the cavability class of the whole ore body is predicted by using the RMR index. The goal of this research is to develop tools for predicting the distribution of cavability class of the block. It is indicated by the grading of ore body cavability in Jinchuan mine area No. 3 that ore rock cavability belongs to extremely apt-medium-sized caving. The result is in conformity with the reality, so this technique can be used in classification of ore rock cavability to guide the real production.