A method is described for studying the thermal shock characteristics of a brittle material. An analysis of the thermal stresses developed in a homogeneous isotropic solid sphere has led to the formulation of an equation relating the physical properties of the body to the temperature difference causing failure and time to maximum stress in a single‐cycle unsteady‐state test. The thermal shock test consisted of plunging a sphere at one uniform temperature into a medium at another temperature. If fracture occurred, the time to fracture was recorded. A large number of tests were run to determine the temperature difference which caused 50% of the spheres to fracture. The thermal shock relationships were tested using a high‐alumina body. The physical properties relating to the thermal shock equations were measured, and calculated temperature differences causing failure and times to maximum stress were compared with measured values. Sufficient agreement was found to lend support to the theory.SummaryA method has been developed for studying the thermal shock characteristics of a brittle substance. The method consists of a single‐cycle test of unsteady‐state nature. Two testing conditions have been selected, one having a rather high surface heat transfer coefficient in a liquid bath and the other having a small finite surface heat transfer coefficient in an air bath. These two conditions are at either extremes regarding the thermal shock usually given a substance in practice.The only fair agreement found in the calculated and experimental data indicates that further investigation is necessary. The importance of certain factors, such as time to maximum stress, which was found to be in rather good agreement for the salt bath heat transfer, cannot be overlooked. There are many applications in the high temperature‐high stress field in which the concept of time to maximum stress might well be examined. For example, when repeated high‐temperature heatings are made on a refractory piece, the cycling might be arranged so that the time to maximum stress was never reached for the particular heating cycle, although the (ΔT) was higher than that necessary to cause failure.Much is to be learned from this type of study which may be applied to actual situations. It must be emphasized, however, that this work has been conducted on one single set of conditions and the factors found here do not necessarily apply to ot