Dual phase (DP) steels are very interesting for lightweight constructions because they combine high plastic deformation potential and high strength. Stress–strain response of multiphase materials similar to DP steel depends on the elastic–plastic behavior of all ingredient phases. In recent years, computational modeling has been successfully established to study the material's mechanical behaviors at microstructure level. In these kinds of modeling the microstructure of dual phase steels can be considered as a matrix of ferrite phase, reinforced by martensite particles. Recent measurements show that mechanical properties of the ferrite phase change with distance from the martensite grains. This phenomenon has not been included in microstructural models precisely until now. In this thesis, a new method is proposed to consider this phenomenon in finite element modeling of dual phase steels microstructure. In this method, ferrite mechanical properties are imported to the model as a continuous function of distance from martensite boundary. For this purpose ferrite mechanical properties are implemented in the model via a user material subroutine (UMAT) in the finite element software Abaqus. In this subroutine mechanical properties are calculated based on the distance from martensite boundaries and previous measurements of the dependency of the properties to this distance. Three types of microstructural models are created based on SEM images. The first model uses a random python code to construct the unit cell in which hexagons martensite grains are randomly dispersed in the ferrite matrix. The second one directly creates the grains and boundaries from the SEM images. The third model uses a Voronoi type algorithm to construct geometries which are statistically similar to the SEM images. In the third model both the crystal plasticity constitutive law and von Mises criterion are employed to model the ferrite and martensite grains while in the others only the latter is used. The tensile test is simulated for the three finite element models in both cases of considering ferrite phase as homogeneous and inhomogeneous matrix. The flow curves of simulations that took the ferrite phase inhomogeneity into account are in better agreement with the experimental flow curves, compared with those of simulations that did not consider it. Considering the ferrite phase inhomogeneity also causes a better prediction of shear band formation in the unit cell compared to the other models. A different stress distribution prediction is also observed in these two models and ferrite phase maximum stress is higher when inhomogeneity is included. These observations can be crucial in investigation of dual phase steels damage. It is observed that martensite volume fraction and the grain size has stronger effect on the model with inhomogeneous ferrite phase that are in better agreement with the experimental results. Keywords: Dual phase steels; Finite element microstructural modeling; Ferrite phase inhomogeneity; Crystal plasticity finite element method