Abstract: According to experimental analysis data of a total of 1918 samples from Baijiantan formation in 446 well area, a study of lithology, porous structure and parameters in connection with reserve estimates was made in this paper. Forming mechanism of seconidary pore system was discussed. The classification and evaluation of reservoirs depend on the reservoir evaluation index(REI). By using multivariate linear regression analysis, the mathematical model to classify and evaluate reservoir with well logging data was established. It is proved that the model is of higher resolving power for identification of low resistivity of sand-detritus reservoir.