A three-dimensional computational fluid dynamics (CFD) dispersion model was developed to simulate odor dispersion from a 3000-sow farrowing farm. The measured odor emission data were used in the CFD model to predict odor concentrations under 30 different meteorological conditions. Atmospheric stability and wind and temperature vertical profiles in the atmosphere were configured in the CFD calculation, and their effects on odor dispersion were evaluated. A Lagrangian discrete phase model driven by a large-eddy simulation (LES) turbulent flow field was presented in the CFD model to predict downwind odor concentration. This Eulerian-Lagrangian approach solved for continuous airflow in the Eulerian reference frame and then solved for trajectories of discrete odor gas parcels (OGPs) in a Lagrangian reference frame. The CFD modeling results were compared with results obtained by the CALPUFF model, a Lagrangian puff model recommended by the U.S. Environmental Protection Agency. The results of both models showed that odor traveled farther under stable conditions than under unstable conditions with the same wind speed. Under the same atmospheric stability category, odor concentrations were higher at lower wind speed than that at greater wind speed. Stronger odor concentration and longer travel distance were favored with stable atmospheric conditions and lower wind speed. Odor concentration results predicted by the CFD model were higher than those predicted by the CALPUFF model in short distances (less than 300 m). Beyond that, CFD predictions were higher than CALPUFF predictions under categories A, B, C, and D, and lower than CALPUFF prediction under category F. The gaps in the odor concentration predictions at longer distances remained stable and were influenced by atmospheric stability category and wind speed.