Characterization of the pore structure of tight sandstone reservoirs carries great significance for the evaluation of the reservoir storage and transport properties, "sweet spots" prediction, and reservoir development. In this paper, the pore structures of tight sandstone reservoirs are characterized on the basis of the multifractal analysis of nuclear magnetic resonance (NMR) transverse relaxation time (T-2) distributions. For the T-2 distribution models established using the mixed Gaussian distribution function, the characteristic parameters of the high probability measure areas (the right branch of the generalized fractal dimension spectrum and the left branch of the singularity spectrum) are closely related to the model parameters (weight coefficients and standard deviations of the short relaxation component). For the T-2 distributions of the tight sandstone samples, the results of the relationships between the multifractal characteristic parameters of the T-2 distributions and the petrophysical parameters of the samples indicate that the characteristic parameters of the high probability measure areas (alpha(max), alpha(0) - alpha(max), D-max, and D-0 - D-max) are closely related to the permeability, T-2 geometric mean (T-2lm), and T-35 (the T-2 value at 35% saturation in the normalized reverse accumulated T-2 distribution curve) values, which can be used to quantitatively evaluate the pore structure heterogeneity of tight sandstones. These samples can be classified into four types on the basis of the shape of the T-2 distribution and the petrophysical parameters. Substantial differences exist among the multifractal characteristics of the different sample types, and several parameters (alpha(max), alpha(0) - alpha(max), D-max, and D-0 - D-max) can be used to classify the sample types. The multifractal characteristic parameters are closely related to the clay mineral content. As the clay mineral content increases, the heterogeneity of the high probability measure areas of the T-2 distribution increases. The results of the multifractal analysis of the NMR logging data further demonstrate the effectiveness of evaluating the pore structure of tight sandstone reservoirs based on the multifractal characteristics of the NMR T-2 distribution.