Abstract: In [1] the author gave a pattern recognition expert system used to analyse the rocky properties of the section image of earthquake surveying, This system includes both the low-level and high-level processing parts. This paper presents a multi-section automatic analysis method for the variation of rocky properties. This method belongs to the low-level processing part, and provides the classification representations of the variation regions of section image rocky properties for the high-level processing part. The method mentioned above can also be used alone, and give the rocky distribution on the vertical section to extract the regions containing oil-gas. At present, the method using the technique of pattern recognition to determine the regions containing oil-gas is realized on single section image, and the performance is not too good, and far from available. The method presented in this paper utilizes multi-section information, hence it has higher value in both theory and practice. This method has been used to process the practical data on CYBER system, and as a result, the picture of the rocky property distribution is presented, and the regions containing oil-gas in the picture are in conformity to the artificial explanation and the location of the oil hole.