•True mode II fracturing is compared to the conventional mode II fracturing.•A criterion is suggested to predict the fracturing type (tensile- or shear-based).•The superiority of the DNBD test over the existing ones is discussed.•True mode II DNBD tests are conducted on three different rock types.•True mode II values of the fracture toughness are compared to the mode I ones. This paper discusses the use of the double-edge notched Brazilian disk test (DNBD) for measuring true mode II fracture toughness of rocks. The term true emphasises that in this test, not only is the crack tip loading shear-based, but also the material failure is shear-induced. Conventional mode II tests typically experience dominantly tensile failure. We introduce a fracture growth criterion that explains where and how a shear-based fracture extension occurs. Our theoretical analysis demonstrates that large values of compressive T-stress in the DNBD specimen significantly help inducing a true mode II fracturing. Crack tip parameters are computed by finite element analyses for various notch lengths and loading angles. These values are then employed to determine the geometry and loading condition for the optimal performance of the test. We also compare the DNBD with two other available tests for measuring true mode II fracture toughness, and show that the DNBD test typically has a lower contribution of mode I loading than the two alternative approaches and, therefore, better approximates the true mode II condition while at the same time being the experimentally simplest. Three types of rocks (limestone, marble and granite) were tested using the new approach and their true mode II fracture toughness is reported for two different crack lengths. The measured true mode II fracture toughness is compared with the mode I fracture toughness obtained from the semi-circular bending test.