The specific problems of DSECs mechanical testing result from the particularities of these 3-D interconnected eutectic ceramics. First of all, 4-point bending tests ensure pure bending loading, whereas 3 PB tests only lead to a tensile and shear stress combination. Consequently, due to the 3-D microstructure of DSECs, interfaces between the various phases are subjected to a mixed (tensile and shear) loading which makes the interpretation of the results (strength) and of the fracture surfaces, rather difficult. For usual ceramics, biaxial flexure testing offers many advantages over 3- or 4-point beam-bending testing. The coaxial-ring test is free of edge influences (flaws): cracks initiate in the central area and propagate outwardly. However, in the case of DSECs, due to the presence of high internal thermal stresses (especially for ternary eutectics), interfaces can be subjected to a strong radial tensile and shear (near the free surface) stress combination. In the presence of the radial tensile stress resulting from biaxial loading, this internal thermal stress combination can lead to premature crack initiation leading to failure. Specimen machining through grinding leads to the formation of a strongly damaged layer. Annealing of this layer leads to the formation of a rough surface: slightly protruding phases and stress concentrations at the interfaces. The measured strength is ae20% lower after annealing than that directly after grinding. Concerning the effect of the microstructure size, four representative sizes have been selected in the ae10 A mu m to submicrometre range. A classical crack propagation criterion has allowed explaining the corresponding strength values.