Abstract: The pore structure of the reservoir rock is directly related to its productivity. The NMR logging data can be used to evaluate the physical properties and pore structure of the reservoir, and then to classify reservoirs and predict productivity.According to characteristics of the tight sandstone gas reservoir of Block A in Ordos basin, the effective porosity, permeability, and pore structure parameters such as separation coefficient, maximum mercury saturation, mean value of pore throat radius and displacement pressure are extracted to create a comprehensive evaluation index. The reservoir classification standard and the productivity predictionmethod are established based on NMR logging data. The NMR prediction productivity is consistent with the tested productivity, which shows that the effect of this method is obvious.