The typical ceramic catalyst support for automotive application has a total volume of 1640 cm3. Approximately 10% of this volume is subjected to tensile thermal stresses due to a radial temperature gradient in service *. These stresses are kept below 50% of the substrate strength to minimize fatigue degradation and to ensure long-term durability . However, the tensile strength measurements are carried out in 4-point bending using 2.5 cm wide x 1.2 cm thick x 10 cm long modulus of rupture bars in which the specimen volume subjected to tensile stress is merely 3.2 cm3 or 0.2% of the total substrate volume . Thus, a large specimen population is often necessary (50 specimens or more) to obtain the strength distribution representative of full substrate. This is particularly true for large frontal area substrates for diesel catalyst supports with an order of magnitude larger stressed volume.In this paper, the modulus of rupture data are obtained as function of specimen size. For example, in addition to the conventional 2.5 cm wide x 1.2 cm thick x 10 cm long specimens, the MOR data for ten different specimen sizes are presented. The stressed volume in these latter specimens is one to two orders of magnitude larger and hence representative of stressed volume in both the automotive and diesel catalyst supports in service. The new data show that the strength values are relatively insensitive to stressed volume indicating that the traditional specimen size for strength measurement is optimum in terms of specimen preparation, specimen testing and converter design. Only a large specimen population is necessary to obtain the complete strength distribution and to arrive at a safe design stress for long-term durability. In this manner, the high temperature fatigue data  based on 2.5 cm wide x 1.2 cm thick x 10 cm long specimens can also be used for estimating the allowable service stress to meet the required converter life. The present study shows that the measured strength can be significantly lower if shear deformation is a significant fraction of total deformation during bending as in the case of thick specimens. In such cases, a correction factor is necessary for obtaining the true strength from measured values.