The relationship between transverse relaxation time (T₂) and pore size distribution is the basis of NMR applications for rocks. However, the equations for T₂ are not accurate enough in rocks with complicated pore structures. Taking pebbly sandstone from the northwestern Junggar Basin in China as an example, the aim of this study is to discover the spatial distribution of pores and its influence on T₂. Porosity, permeability, micro-images and T₂ distributions were acquired from rock samples, and pore structure parameters were obtained from binarized thin section images. The results show that as the grain size increases, the proportion of dissolution pores increases and the spatial distribution of pores changes from a random to a clustered pattern. The relaxation of a hydrogen atom takes longer and T₂ is higher in dissolution pores compared with those in intergranular pores. New equations for T₂ that consider the spatial distribution of pores are proposed.