Assessing quantitatively the microscopic pore structures of porous rocks, including irregularities of pore shapes and pore size distributions, is becoming one of the most challenging efforts. Nuclear magnetic resonance (NMR) measurements were used to provide insights into the pore geometry (pore size and shape) and pore connectivity of the Chang 7 tight shaly sandstones (in situ permeability <0.1 md) in Ordos basin. The incremental transverse relaxation time (T2) distributions for the 100% brine-saturated samples display unimodal and bimodal behaviors, presenting a geometrical arrangement composed of small to large pore size domains. The NMR parameters such as bulk volume irreducible (BVI) (capillary and clay-bound water), free fluid index (FFI) (movable water), the value of T2 separating the BVI from FFI, the amplitude weighted mean on a logarithmic scale (T2gm), and the value of T2 that shows the highest frequency on the T2 spectrum (T2peak) for each sample were determined. Then the fractal theory was adopted to quantitatively express the complexity and heterogeneity of the sandstones. The results show that only minor primary intergranular porosity remains, and variable amounts of micropores and secondary intragranular porosity with poor connectivity occur in the Chang 7 tight shaly sandstones. Assuming spherical pores, a new model to calculate the fractal dimension of pore structure from the NMR T2 distributions is proposed. The fractal dimensions of all the samples are calculated, and the accuracy of the proposed model is verified by the regression coefficients. The microscopic pore structures are heterogeneous in these tight sandstones according to the high value of fractal dimensions. Micropores are the primary causes of heterogeneity in tight sandstones, and samples with unimodal T2 distribution behaviors and high content of short components have the highest fractal dimension and heterogeneity. The calculated fractal dimension is strongly correlated with T2peak and T2gm; therefore, the fractal model proposed in this study can be used to calculate the fractal dimensions and evaluate the heterogeneities of the porous rocks satisfactorily. The fractal model proposed in this study helps to quantitatively assess the pore structures of tight sandstones using NMR measurements.