By using different permeability prediction models based on petrologic analysis of the sandstone, the permeability prediction of sandstone was studied for the Upper Shaximiao formation in Xinchang area of Western Sichuan Basin. The re- suits show that the sandstones are secondary porosity dominated with poor porosity-permeability relationships and consequently have a disparity between gas production and reservoir evaluation using porosity data. The reservoirs were classified on the ba- sis of component, texture and pore space types of sandstones, and the corresponding permeability prediction models of porosi- ty-permeability relationship were developed, which can increase the correlation between porosity and permeability. The reser- voir types can be differentiated by analyzing the relationships between reservoir characteristics and well log parameters. Based on enhanced well logging porosity interpretation, more accurate permeability values are obtained by using well-log-calculated porosities and permeability prediction models, which is helpful for improving reservoir evaluation.