Abstract: The method of three-dimensional seismic exploration for salt rock cavern is studied and analyzed. Through the example seismic exploration data, the sensitive seismic attributes of salt rock cavern are selected, and the seismic multi-attribute analysis is carried out, and four sensitive seismic attributes of salt rock cavern are optimized, the four seismic attributes are optimized by Principal Component Analysis（PCA）. The results show that after PCA analysis, the redundancy among various data is eliminated, and the transformed principal components are transformed, geophysics and geological information that represents multiple attributes prior to principal component analysis can better highlight the local characteristics of the signal. The optimized results are applied to the prediction of rock salt cavity, and a good result is obtained.