AbstractDespite substantial clinical interest in the fracture resistance of human dentin, there is little mechanistic information in archival literature that can be usefully used to model such fracture. In fact, although the fracture event in dentin, akin to other mineralized tissues like bone, is widely believed to be locally strain‐controlled, there has never been any scientific proof to support this belief. The present study seeks to address this issue through the use of a novel set of in vitro experiments in Hanks' balanced salt solution involving a double‐notched bend test geometry, which is designed to discern whether the critical failure events involved in the onset of fracture are locally stress‐ or strain‐controlled. Such experiments are further used to characterize the notion of “plasticity” in dentin and the interaction of cracks with the salient microstructural features. It is observed that fracture in dentin is indeed locally strain‐controlled and that the presence of dentinal tubules does not substantially affect this process of crack initiation and growth. The results presented are believed to be critical steps in the development of a micromechanical model for the fracture of human dentin that takes into consideration the influence of both the microstructure and the local failure mode. © 2003 Wiley Periodicals, Inc. J Biomed Mater Res 67A: 484–495, 2003