Abstract Pore network modeling and physical experiments were employed to understand the migration and accumulation of tight oil in the Lucaogou Formation of the Jimsar Sag, Junggar Basin, China. The pore structure characteristics of tight rock samples were statistically analyzed, from which some representative network models were constructed. The estimated petrophysical properties of these networks, e.g., porosity, permeability, and pore/throat size distribution, agree well with those obtained from experiments, which justified the effectiveness of our models. The experiments and simulations of oil migration into tight rocks suggest that there are mainly two different types of accumulation patterns, i.e., rapid growth pattern and consistent growth pattern, which are characterized by S-shaped and modified S-shaped correlations of oil saturation and pressure gradient, respectively. Oil accumulation mainly occurs in the middle stage; however, the increment of oil saturation of consistent growth pattern is much slower than that of rapid growth pattern. The water-oil relative permeability curves, which describe the fluid transport capabilities during oil migration, were also computed from simulations. The effects of coordination number, aspect ratio, as well as clay volume were analyzed. Results show that a core sample having better connectivity and more homogeneous pores and throats facilitates the oil migration; however, a higher clay volume impedes oil accumulation and decreases the final oil saturation. This work not only provides a better understanding of tight oil accumulation mechanisms, but also supplies an alternative avenue to study multiphase flow in tight rocks.