Polymer membranes are ubiquitously utilized in a wide variety of technological applications; thus, the characterization and fundamental understanding of their properties in relevant working environment are indispensable. In this study, we develop a methodology to computationally characterize the microscopic structures and properties of polymer membranes in liquids by integrating atomistic simulations and two characteristic lengths; the characteristic lengths are proposed from the concept of Gibbs dividing surface. The methodology is illustrated for the swelling of an ultrathin polymer of intrinsic microporosity (PIM-1) membrane in different solvents (acetone, acetonitrile, methanol, ethanol, and water). Specifically, the swelling dynamics is examined by time-dependent solvent content and radius of gyration in the membrane inner layer. The simulated swelling factors are found to agree well with-available experimental data. In the five solvents, the predicted swelling degrees follow the experimentally measured trend. Quantitative relationships are observed for the swelling degree with the Hildebrand solubility parameter and the polymer-solvent interaction energy. Furthermore, the swelling of PIM-1 membrane in seawater is simulated; the ions are completely rejected during swelling, and the swollen membrane is structurally similar to the one in pure water. The computational methodology developed in this study can be applied to other polymer membranes in various solvents and liquids, thereby expanding the scope of potential applications.