The conceptual and computational issues regarding the development of models to predict microstructure/ mechanical-property relationships in advanced ceramics are discussed. Advanced ceramics provide a particular challenge because their higher toughness or reliability, relative to monolithic ceramics, is a result of stable, distributed damage evolution. Capturing the details of distributed damage at atomistic and microstructural length scales is computationally prohibitive, but only in certain systems does it appear plausible to neglect atomistic and crack-tip details in favor of larger-scale damage propagation and interaction. Examples, mainly from the author's own work, are briefly presented to indicate the range of problems that have been addressed and the relative successes and failures. © 1996 ESCOM Science Publishers B.V.