Ductile fracture in steels is essentially due to cavity nucleation, growth and coalescence. Once voids are initiated in the metallic matrix, and the deformation continues to increase, voids grow in size and finally coalescence to fracture. This dissertation investigates ductile fracture initiating in the bulk of material for relatively low stress triaxialities. Particularly zero stress triaxiality conditions, i.e. under shear dominated loading, provided for an isolated study of single mechanisms such as void nucleation, and shear localization. In this dissertation, we show that the idealized division of ductile fracture in a sequence of elementary processes is not appropriate for the selected quenched and tempered VAR steel. Ductile fracture is nucleation-controlled, i.e without meaningful void growth. Void nucleation occurs by debonding of tempering carbides from the surrounding metallic matrix. The carbide-matrix interface strength is estimated by a dislocation-based stress decohesion model. By means of FEM computations with a unit cell simulations comprising a single carbide, the role of the carbide-matrix interface strength is emphasized. The results of FE computations demonstrate that the carbide-matrix interface strength is the key parameter controlling damage initiation and its evolution to fracture.