Increasing demands for lightweight construction make the use of press-hardened steels (PHS) more and more important. In addition to a high strength, it is important for the material to guarantee a sufficient resistance against the formation and propagation of cracks. The crash performance of the material is tested technologically by a special three-point bending test where a critical bending angle is determined. To improve the crash performance of PHS, the damage evolution during bending was investigated in the present work with the help of interrupted bending tests. The results of these studies show the following failure mechanism: In a first step, small surface cracks and notches appear on the flattened surface of the specimen opposite to the loading pin. Simultaneously, shear bands are formed at these positions, emanating from the surface. The final failure occurs due to crack formation in the vicinity of the shear bands. Additional interrupted three-point bending tests on other high-strength steels resulted in the appearance of two different failure mechanisms, which are determined by the type of the microstructure and the magnitude of the strain hardening exponent. It was found that only materials with tempered martensite in the microstructure and small hardening exponent, n2-4 ≈ 0.06, show the failure mechanism that has been observed in PHS. Furthermore, it was demonstrated that the exact condition of the microstructure plays an important role for the bending ability of PHS. Tempering improves the critical bending angle, but only at the cost of a decreased tensile strength. A possibility for distinct improvement of the bending ability without decrease in strength was found by decarbonisation of the sample surface. It was also tried to optimize the influence of the chemical composition for PHS in order to improve the mechanical properties. For that purpose, new laboratory-scale experimental alloys were cast, in which the amounts of aluminium, manganese, manganese + molybdenum, as well as manganese + molybdenum + niobium were alternatively increased. The first and the last of the aforementioned types seem to have a potential for further improvement and, therefore, has been industrially produced and further developed.