This paper explores the applicability of nuclear magnetic resonance (NMR) technology on pore size distribution (PSD) and movable fluid distribution (MFD) in tight sandstones. Centrifugation experiments and NMR tests are performed on saturated samples. The fluid changes in pores corresponding with three different types of NMR T-2 distribution after each centrifugation is then analyzed. In addition, a new method to determine the conversion factor from NMR T-2 distribution to PSD is developed. In comparison with the PSD obtained by mercury intrusion porosimetry, the new method is more suitable for PSD calculation in tight sandstones. Afterward, the optimum centrifugal force to determine the threshold radius for fluid flow is obtained. On the basis of this, we analyze MFD in tight formation. Through study, the following results are arrived at: patterns of NMR T-2 distributions of outcrop and subsurface cores at water saturation condition can be classified into three types (I, II, and III). Among which, type I and type II show a better pore connectivity than type III with NMR T-2 distribution of a higher movable peak and a lower immovable peak. The optimum centrifugal force for the Chang 6 tight formation to determine movable fluid is 418 psi and pores show no obvious difference with throats when radii are less than 0.05 mu m. Movable fluids are mostly controlled by throats with radii smaller than 1 mu m, especially throats with radii between 0.3 and 1 mu m. Movable fluids are mostly stored in pores around the movable peak of bimodal NMR T-2 distribution with radii ranging from 10 to 100 mu m. These pores are residual interparticle pores and dissolution pores. The sets of experiments and the new method presented in this paper are proved effective in quantitively describing PSD and also qualitatively evaluating pore throat connectivity in tight sandstones. Petro physical characterization by NMR technique provides an effective approach to better understand pore throat structures and storage capacity of tight oil reservoirs.