Permeability, irreducible water saturation and other information associated with the pore size, such as capillary pressure, are very important input parameters in tight gas reservoir evaluation. Fortunately, those parameters can be well estimated from field NMR logs, if the proper calibrations between NMR relaxation time and capillary pressure were established with measurements from core samples. In this study, a novel calibration technique, which is used for constructing capillary pressure curve from NMR logs by combining the classical SDR model with J function, is proposed and the corresponding model is established. Based on the laboratory measurements of 54 plug samples taken from the Upper Triassic sandstone formation (Xujiahe Formation) in Anlu district, Sichuan basin, Southwest China, the model parameters are calibrated. With the calibrations established and procedures proposed in this paper, the synthetic capillary pressure curves of reservoir can be constructed from field NMR spectrum. Also from those laboratory measurements, the parameter of Kϕ (defined as the pore structure index) can be well related to the Swanson's parameter derived from capillary pressure curve. This correlation could then be used for estimating reservoir permeability from the synthetic capillary pressure curve. After formation is classified into three types by using the pore structure index, the pore throat radius cutoffs can be acquired to calculate irreducible water saturation from the pore throat radius distribution. The proposed techniques and procedures are applied in field logs and the estimated information and parameters are well verified by core and drill stem testing data. By combining with the information of reservoir pore structure, permeability and irreducible water saturation, potential low resistivity contrast gas bearing reservoirs can be identified from water layers.