Glutenite reservoir has complicated pore structure and strong heterogeneity and traditional logging interpretation methods often have nonignorable calculation errors in permeability evaluation. Thus, there is no classic model to calculate the permeability of glutenite. This Paper takes Badaowan group formation of Karamay Oilfield as an study example. Firstly, the main controlling factors of permeability were analyzed at the micro level. Secondly, three sets of permeability logging interpretation methods were built according to the study area’s situation: the first is the improved multivariate regression model based on the predecessors’ research; the second is the different permeability models of different lithology based on lithology identification; the third is the BP neural network. Finally the verification results showed that compared with the traditional empirical formula and the multivariate regression model, permeability model based on different lithology and the BP neural network had better effects in the practical application, with significant improvements in the precision of logging interpretation and better application prospects in strong heterogenous glutenite reservoir.