Abstract: Traditional NMR logging permeability models are always poor in application effect to tight sandstone reservoir. Considering the affects of formation pore?throat structure, pore space distribution, pore specific surface area on tight sandstone reservoir permeability, this paper proposes NMR logging permeability prediction method based on dynamic pore structure efficiency by means of theory of equivalent rock composition, and presents the model using29rock samples. In order to improve the prediction accuracy of the irreducible water saturation and the pore structure efficiency in this model, it gives such idea for evaluation of these two parameters as using T2 geometric mean value and4~32ms NMR porosity to build BP neural network for point?by?point prediction T2 cutoff value, and then taking the T2 cutoff value as intermediate parameter to get the irreducible water saturation and the porosity structure efficiency. The handling of the data proves that this model is of higher calculation accuracy than SDR and Coates models, and can be applied to permeability calculation of tight sandstone reservoir, and helpful to the productivity evaluation of the reservoirs