In order to build a model for the Chang-8 low permeability sandstone reservoir in the Yanchang formation of the Xifeng oil field,we studied sedimentation and diagenesis of sandstone and analyzed major factors controlling this low permeability reservoir.By doing so,we have made clear that the spatial distribution of reservoir attribute parameters is controlled by the spatial distribution of various kinds of sandstone bodies.By taking advantage of many coring wells and high quality logging data,we used regression analysis for a single well with geological conditions as constraints,to build the interpretation model for logging data and to calculate attribute parameters for a single well,which ensured accuracy of the 1-D vertical model.On this basis,we built a litho-facies model to replace the sedimentary facies model.In addition,we also built a porosity model by using a sequential Gaussian simulation with the lithofacies model as the constraint.In the end,we built a permeability model by using Markov-Bayes simula-tion,with the porosity attribute as the covariate.The results show that the permeability model reflects very well the relative differences between low permeability values,which is of great importance for locating high permeability zones and forecasting zones favorable for exploration and exploitation. In order to build a model for the Chang-8 low permeability sandstone reservoir in the Yanchang formation of the Xifeng oil field, we studied sedimentation and diagenesis of sandstone and analyzed major factors controlling this low permeability reservoir. By doing so, we have made clear that the spatial distribution of reservoir attribute parameters is controlled by the spatial distribution of various kinds of sandstone bodies. By taking advantage of many coring wells and high quality logging data, we used regression analysis for a single well with geological conditions as constraints, to build the interpretation model for logging data and to calculate attribute parameters for a single well, which has accuracy of the 1-D vertical model. On this basis, we built a litho-facies model to replace the sedimentary facies model. built a porosity model by using a sequential Gaussian simulation with the lithofacies model as the constraint. In the end, we built a permeability model by using Markov-Bayesian simula- tion, with the porosity attribute as the covariate. The results show that the permeability model reflects very well the relative differences between low permeability values, which is of great importance for locating high permeability zones and forecasting zones favorable for exploration and exploitation.