One of the main challenging issues in friction stir welding (FSW) of stiffened structures is maximizing skin and flange mixing. Among the various parameters in FSW that can affect the quality of mixing between skin and flange is tool plunge depth (TPD). In this research, the effects of TPD during FSW of an Al-Mg-Si alloy T-joint are investigated. The computational fluid dynamics (CFD) method can help understand TPD effects on FSW of the T-joint structure. For this reason, the CFD method is employed in the simulation of heat generation, heat distribution, material flow, and defect formation during welding processes at various TPD. CFD is a powerful method that can simulate phenomena during the mixing of flange and skin that are hard to assess experimentally. For the evaluation of FSW joints, macrostructure visualization is carried out. Simulation results showed that at higher TPD, more frictional heat is generated and causes the formation of a bigger stir zone. The temperature distribution is antisymmetric to the welding line, and the concentration of heat on the advancing side (AS) is more than the retreating side (RS). Simulation results from viscosity changes and material velocity study on the stir zone indicated that the possibility of the formation of a tunnel defect on the skin–flange interface at the RS is very high. Material flow and defect formation are very sensitive to TPD. Low TPD creates internal defects with incomplete mixing of skin and flange, and high TPD forms surface flash. Higher TPD increases frictional heat and axial force that diminish the mixing of skin and flange in this joint. The optimum TPD was selected due to the best materials flow and final mechanical properties of joints.