After long-term production by water injection, the conductivity features of oil reservoirs are changed, leading to difficulties in calculating remaining oil saturation by traditional well-logging techniques. There are some specific phenomena relating to petrophysical properties in water-flooded layers, and a deep investigation of them could help us understand resistivity changes of rocks during the water-injection process. We compared traditional mixed-fluid resistivity methods and determined their drawbacks. In addition, we revised the theoretical models and came up with a novel method to calculate mixed-fluid resistivity. The displacement process was divided into three stages: non-water-flooded stage, low and medium water-flooded stage, and high water-flooded stage. To study the relationships between rock resistivities and water saturations when injecting fresh-/saltwater, we used digital rock numerical methods. We first constructed 3D digital rock models based on the process-based method, and then we obtained oil and water distributions in the pore space by the pore morphology method. When resistivities of every component in digital rock models were assigned, rock resistivities could be calculated in different water saturations by the finite-element method. We found that in the case of freshwater injection, the relationship between water saturations and rock resistivities formed an S-shaped curve, but in the case of saltwater injection, if the injection water salinity was higher, the rock resistivity would be decreased by increasing the water saturation. In different rock heterogeneity and wettability conditions, these curves showed similar shapes and trends. Combined with resistivity logging data, this novel approach was used in the interpretation of a real oil well. Compared with the existing models in the interpretation practices, this new model better matched the real-field saturations and improved the accuracy of saturation interpretation in the studied water-flooded oil reservoir.