Nowadays, materials, structures and devices can be designed and fabricated at nanoscale dimensions with controlled microstructures processed “atom by atom”. Predicting their toughness and ensuring their durability over the long term still present a major challenge. These properties esult from the complex interplay between the applied stress conditions at the macro scale and the material heterogeneities at the scale of the microstructure, crucially influenced by processes such s damage, chemical reactions and thermal fluctuations, also occurring at the microscale. Bridging length scales is central in the study of material failure and the main objective of my esearch project is (i) to provide a statistical and quantitative description of the effect of material microstructure on the propagation of cracks and the damage spreading in highly heterogeneous brittle and quasi brittle solids. This will be achieved by investigating experimentally and numerically at the microscale the failure of model systems with controlled heterogeneous properties; (ii) to implement these mechanisms into simplified predictive models based on tools issued both from statistical physics and fracture mechanics. They will link material characteristic nd failure mechanisms at the microstructure scale with their macroscopic failure properties. In the long run, the models developed within this project will help optimizing efficiently the microstructure f a solid in order to achieve improved failure properties, such as strength and lifetime.