Abstract It is well known that the measured values of the fracture toughness of rocks are influenced by material heterogeneity, dimensions, boundary conditions, and asymmetric mechanical behavior. Consequently, the results obtained by standard testing methods developed primarily for homogenous materials with symmetric mechanical behavior, can significantly differ. The standard methods take global approach. Thus, they suppose that the material tested will follow a specific physical model and that one can consider the selected testing method as a black box in which some simple characteristics are measured and the required values can be evaluated. If the material behavior is too different from the theoretically expected one, this global approach will fail. The authors present a method called Local Fracture Toughness Testing (LFTT) to overcome these obstacles. LFFT is calculated independently of the boundary conditions and the crack length. LFTT is based on a complex methodology using a series of tomographic reconstructions, for which data are recorded during specimen loading. Subsequent extended data processing using digital image correlation serves for calculating the evolution of the displacement/strain fields and for identifying the crack which develops during increased loading. Later on, the crack tip opening displacement and the local fracture toughness KIC are calculated at arbitrarily selected positions independent of the geometry and boundary conditions. The LFTT methodology was tested on a sandstone specimen, since such material is usually considered to be brittle. In this work, the authors demonstrate that even a stable crack extension can be identified after maximal loading. Using a loading machine developed in-house, the experimental data allowed for the measurement of fracture toughness at five loading levels/crack lengths. In addition, fracture toughness was measured in nine planes crossing the crack tip for each loading level.