•Fracture tests are performed on U-notched SGFR-PA6 specimens under mode I loading.•The Fictitious Material Concept is combined with TCD for critical load prediction.•FMC-TCD criterion is found successful, but the original EMC-TCD criterion is not.•TCD is used in its basic form without the need for critical distance calibration. In the present research, the fracture prediction of U-notched single edge notched bend (SENB) specimens made of the short glass fibre reinforced polyamide 6 (SGFR-PA6) with variable moisture is investigated experimentally and theoretically. In the experimental program, numerous rectangular specimens weakened by edge U-shaped notches of different tip radii are tested for fracture under pure mode I loading, and the load-carrying capacity (LCC) of the specimens is experimentally measured. In the theoretical program, in order to avoid time-consuming and complex elastic-plastic failure analysis, and due to the significant strain-hardening and the substantial strain-to-failure of the composite material tested, the prediction of the experimentally obtained LCCs is approached in order to test its validity using the Fictitious Material Concept (FMC), proposed most recently by the first author, in combination with the well-known Theory of Critical Distances (TCD). According to the new proposed failure model, since the fictitious linear-elastic material is used, TCD can be applied by means of its basic formulation without considering any previous calibration of the corresponding critical distance. It is revealed that the FMC-TCD combined criterion can predict the experimental results well.