This study presents a systematic modeling approach for examining the efficiency of the MEOR process based on in situ selective plugging by bacterial biopolymer production and optimization of the nutrient injection strategy to yield the maximum oil recovery. This study focuses on modeling in situ selective plugging by the bacterial biopolymer dextran that is generated by Leuconostoc mesenteroides. Bacterial growth and dextran generation were described by a stoichiometric equation and kinetic reactions using batch model simulation. Based on the parameters for permeability reduction obtained from the sandpack model, the MEOR process was implemented in a pilot-scale system that included a highly permeable thief zone in a low-permeability reservoir. The base MEOR design yielded a 61.5% improvement of the recovery factor compared to that obtained with waterflooding. The parametric simulations revealed that the recovery efficiency was influenced by the amount of dextran, as well as the distribution of dextran, and thus, the injection strategy is critical for controlling the dextran distribution. By incorporating the results from the sensitivity analysis and optimization to determine the optimal design parameters, a 36.7% improvement of the oil recovery was achieved with the optimized MEOR process in comparison with the base case.