The feasibility of applying the fracture mechanics is explored. Discrepancies in measured values due to different specimen shapes, different experimental methods and whether sawed notches or natural cracks have been used show that it is problematic to compare results. Starting from the concept that strength and fracture behavior are largely governed by microstructure and microcrack processes, an attempt is made to explain deviations from the behavior predicted by fracture-mechanical considerations and to define limits of application.