For some years ceramic bearing balls based on silicon nitride have been routinely used in technical practice. An important property of bearing balls is their strength, but appropriate testing methods are still missing. In this paper four different methods for strength testing are applied to commercial bearing balls. Each of the tests needs a different type of specimen, their preparation needs a very different effort, and the stress state applied to the specimens is also very different. This causes pros and cons, which are discussed in detail. The conventional 4-point bending test characterises the material in the interior of the balls. The applied stress state is uniaxial. The machining of the bending bars out of the balls is time intensive and costly. The ball on three balls test also characterises interior of the balls. The stress state is biaxial. The machining of the disc shaped specimens out of the balls is less expensive than the production of bending bars, but the finish of the tensile loaded surface needs special care. The data of both types of tests can be converted into each other using Weibull theory. The specimens in the triple ball crush test are as-received bearing balls, which are squeezed together. This causes some kind of contact loading, as will also occur in service. Failure is caused by the creation and growth of contact cracks, followed by a collapse of the compressed and cracked material. A detailed analysis of test results is complicated. It can be speculated that the component’s behaviour is mainly influenced by the toughness of the material and that the flaws in the material or at the component’s surface are of less significance. In the newly developed notched ball test the highest stressed region is a part of the original surface of the balls. Machining of the notch is straightforward. The stress state is almost uniaxial. The strength depends on size of flaws in the surface region. Therefore the notched ball test is a relevant measure to characterize the quality of the bearing balls.