The mechanical strength of polycrystalline silicon is discussed in terms of activation of critical flaws, as well as material microstructure and inhomogeneity. The Weibull probability density function parameters were obtained to deduce the scaling of material and component strength and to identify critical flaw populations, especially when two or more flaw types are concurrently active. It was shown that scaling of strength can change for self-similar micronsized features, which limits the applicability of strength data from large MEMS components to small MEMS components. On the other hand, the probability of failure for small components is described by a larger Weibull material stress parameter, which makes uniaxial strength data appropriate for conservative design. Furthermore, according to mode I and mixed mode I/II fracture studies for polysilicon, it is concluded that variation in the local critical energy release rate, owed to microstructural inhomogeneity, accounts for up to 50% scatter in strength (with reference to the minimum recorded value.) Thus, the conditions for the applicability of the Weibull probability density function in polycrystalline silicon are rather weak, because flaws of the same length that are subjected to the same macroscopic stresses are not always critical.