The determination of pure shear (Mode Ⅱ) fracture toughness,KIIc,for brittle rock is studied. Three types of specimens,which are anti-symmetric four-point bending specimen,direct shearing specimen and shear-box specimen,are experimentally investigated. A new criterion is established for either Mode I (tensile) or Mode Ⅱ (shear) fracture of brittle rock,which clearly distinguishes the fracture mode from the loading mode. With guiding-grooves in both the anti-symmetric four-point bending and the direct shearing specimens,Mode Ⅱ fracture can be successfully created. Numerical calculations show that the guiding-grooves in the notch plane can depress the tensile stresses at the notch tips and enable the crack to propagate along the notch plane. The value of KIIc obtained in this case is 2~3 times as high as the magnitude of KIc,which is considered as the true Mode Ⅱ fracture toughness of rock. Mode Ⅱ fracture can be easily created in a shear-box loaded square specimen.The compressive stress in the two directions can completely depress the tensile stresses around the notch tips and makes it easy for the crack to initiate and propagate along the notch plane,leading to Mode Ⅱ fracture. The shear-box test is suggested as a method for determining the Mode Ⅱ fracture toughness of rock. Both theoretical and laboratory studies of shear-box test show that the apparent Mode Ⅱ fracture toughness ,i.e. Mode Ⅱ fracture toughness under the compressive stress,linearly increases with the compressive stress on the notch plane. The intercept of straight line of versus (i. e. = 0) is defined as the Mode II fracture toughness KIIc. In some cases,it is not KIIc but that needs to be known. For instance,the pre-existing cracks and discontinuities in rock and rock masses are subjected to compressive loading in most cases. What is useful in this case is the Model II fracture toughness under the in-situ compressive stress, ,instead of KIIc. The size effect on the determination of KIIc has been experimentally studied. The examined dimensions include the specimen thickness B,the dimensionless notch length a /W,and the notch inclination angle a. Test results show that KIIc decreases with the increment of the specimen thickness B and becomes a constant when B is equal to or larger than the specimen height W for both the single- and double-notched specimens. When the dimensionless notch length a /W increases,KIIc decreases and approaches a limit if a /W is in the interval of 0.5~0.6 for single-notched specimens and 0.3~0.35 for double-notched specimens. The notch inclination angle has a significant effect on the failure pattern of the specimen. Too small or too large a will lead to either local crushing or Model I fracture. The favorable range for a is 60°~75° for Model Ⅱ fracture. Finally,a new criterion is established based on the examination of the Mode Ⅰ and Model Ⅱ stress intensity factors,KI(q ) and KII(q ). The criterion can predict either Model I or Model Ⅱ fractures,no matter what kind of loading condition is applied. Model Ⅱ fracture occurs when the ratio of KIImax to KImax is larger than that of KIIc to KIc and KIImax reaches its critical value,KIIc.