For large-scale engineering applications, macroscopic phenomenological damage mechanics models with less complexity are usually applied due to their high computational efficiency and simple implementation procedures in finite element simulations. Compared with micromechanical models, however, they also have a significant disadvantage, namely the lack of microstructure sensitivity. This work aims to develop a method to integrate the influence of microstructural features into the parameter calibration of a stress-state-dependent damage mechanics model (the modified Bai-Wierzbicki model) for a C45E + N steel. For this purpose, virtual experiments are performed on an artificial microstructure model to derive the plasticity and damage initiation behaviour for the investigated material. A crystal plasticity model for ferrite along with an empirical strain hardening law for pearlite are assigned to the corresponding constituents in the artificial microstructure model to define their material properties. Nanoindentation tests and numerical analysis are used to calibrate the parameters of the crystal plasticity model. By applying different boundary conditions to the artificial microstructure model, both the plasticity and the damage initiation behaviour under different stress states are calibrated by the virtual experiments. In addition, this approach is also applied to investigating the influence of microstructure on plasticity and damage initiation.