Soybean (Glycine max (L.) Merrill) may symbiotically fix large amounts of atmospheric nitrogen (N). However, crop growth, uptake of soil mineral N, and N fixation are sensitive to crop growth stage, weather, and soil water and mineral N contents. Accurate simulation of soybean growth, N uptake and fixation, and soil N contents require that those factors be considered. Erosion-Productivity Impact Calculator (EPIC) is a mathematical model capable of simulating complex crop rotations over many years. In order to accurately simulate the beneficial effects of legumes in crop rotations, a model of nitrogen fixation is needed. This paper describes the development and testing of the soybean N-fixation submodel in EPIC. The functional relationships in the N-fixation submodel were derived from the literature. Data from a long-term field experiment were used to calibrate submodel sensitivity to soil water contents and its simulation of crop N contents and fixation. The model was validated with independent data from an experiment in which soil depth, irrigation, and nitrogen fertilizer were varied to produce a wide range of crop yields and N fixation. Relatively good agreement was found between simulated and measured values of biological N fixation, N uptake from the soil, and crop biomass and yields.