The performance of materials such as steels, their high strength and formability, is based on an impressive variety of competing mechanisms on the microscopic/atomic scale (e.g. dislocation gliding, solid solution hardening, mechanical twinning or structural phase transformations). Whereas many of the currently available concepts to describe these mechanisms are based on empirical and experimental data, it becomes more and more apparent that further improvement of materials needs to be based on a more fundamental level. Recent progress for methods based on density functional theory (DFT) now makes the exploration of chemical trends, the determination of parameters for phenomenological models and the identification of new routes for the optimization of steel properties feasible. A major challenge in applying these methods to a true materials design is, however, the inclusion of temperature-driven effects on the desired properties. Therefore, a large range of computational tools has been developed in order to improve the capability and accuracy of first-principles methods in determining free energies. These combine electronic, vibrational and magnetic effects as well as structural defects in an integrated approach. Based on these simulation tools, one is now able to successfully predict mechanical and thermodynamic properties of metals with a hitherto not achievable accuracy.