Adaptive remeshing finite element model is used for the prediction of material behavior and temperature histories in friction stir welding (FSW). Then, Monte Carlo method with nucleation in every MC step is used for the simulation of the grain growth in FSW. In addition, the process of dynamic recrystallization which results in the nucleation of new grains has been modeled. The model is validated by comparison of both experiments and numerical results from the literature. The width of the stirring zone has been estimated by monitoring the movements of the traced material particles and by identifying the region containing dynamically recrystallized fine grains. The inner border of the heat-affected zone has been determined by monitoring the material flow, while the outer border has been identified by monitoring the average grain size as a function of the lateral distance from the weld. The effect of two FSW process parameters, tool rotational speed and tool shoulder diameter, is investigated. An increase in either of these two parameters has been found to increase the overall grain size as well as the width of the welding zones.