In many industrial applications, combination of compressive and shear loads will act on refractory linings. Therefore the prediction of lining failure requires the knowledge of the multiaxial behaviour of the refractory materials under service conditions. In order to take into account those aspects in modelling, Drucker-Prager criterion is often used to describe the mechanical behaviour of granular materials. It applies a linear dependency of the shear strength on the hydrostatic pressure. Therefore the Drucker-Prager failure line requires the knowledge of two essential parameters which should be determined experimentally: the cohesion (d) representing the failure shear stress without any hydrostatic pressure, and the friction angle (β) defining the increase of the failure shear stress with hydrostatic pressure. For several materials experimental data are available in literature, especially in the field of geology or civil engineering. But, unfortunately characterization techniques applied so far are operating at ambient temperature only. For refractories, cohesion and friction angle have also to be determined at elevated temperature (up to e.g. 1500°C). A simple adaptation of available experimental devices, developed for room temperature measurements, to elevated temperatures is not possible. The present work proposes a new approach to carry out such measurements in the case of refractory materials. The main idea is to apply an uniaxial load on a bar shaped specimens with an inclined notch in order to obtain locally a combination of shear- and compressivestresses. It has the advantage to necessitate the facilities of a compression test only which is rather easy to perform even at elevated temperatures. With this testing procedure the determination of d and β values requires at least two specimen designs with different notch angle. Specimen design has been optimized using FEM modelling. In a first step, the comparison of d and β values measured at room temperature with values obtained by a classical triaxial test allows to validate this new approach. In a second step, values have also been successfully measured at elevated temperatures for two different refractory materials. A second geometry has been optimized using FEM modelling in order to use a torsion-compression testing device. Here the specimen is loaded uniaxially and subjected to torsion and can also be tested at elevated temperature. The first values at room temperature have been validated.