The T(2)cutoff is an important input parameter in nuclear magnetic resonance (NMR) applications. The accuracy of the T(2)cutoff affects the prediction reliability of parameters associated with the identification and evaluation of formations. Current methods are based on regional statistics; they have limited applications and the predicted values are not always reliable. In this study, a total of 36 core samples, drilled from Triassic tight sandstone reservoirs of the southwest Ordos Basin in China, were used for laboratory NMR measurements under fully brine saturated and irreducible water conditions. Based on the morphological character analysis of experimental NMR T-2 spectra, we demonstrate that NMR T-2 spectra can be fully simulated using the normal distribution function. In addition, a new method was proposed, which can predict various T(2)cutoffs based on the morphological differences of NMR T-2 spectra, and the irreducible water saturation (S-wirr) was calculated, which represents the ratio of the sum volume of clay bound water and capillary bound water to total pore volume. The reliability of this method was verified by comparing the predicted T(2)cutoff and S-wirr values with those of core NMR experimental results. Finally, we extended this method into field applications in several tight sandstone reservoirs in China. The results show that total T-2 distributions of the formation were fully simulated by the normal distribution function, and various T(2)cutoff and S-wirr values were precisely predicted. Meanwhile, tight sands permeability curves, predicted based on the Timur-Coates model, were also accurately estimated. The results of our study may be applied to improve tight sandstone reservoir identification and evaluation using NMR logs.