From July 10th through July 13th, 1994, an informal workshop co-organized by RILEM committees 116-PCD and 123-MME was held at Saint-Remy-Ies Chevreuse, France, and attended by 38 delegates from 16 countries. Twenty-nine papers were presented, converging the general subjects of modelling micro structures and predicting durability of concrete and other cement-based materials. A short summary follows: G. M. Idom's paper entitled "Modelling Research for Concrete Engineering" serves as an introduction to the workshop, presenting an overview of modelling research with the conelusion that the broad practica1 objective is to produce high-quality concrete. This means that many characteristics, ranging from rheology to alkali-silica reaction, must be modelled. In other words, the system must be understood. Idom's paper sets the stage for papers in two general areas: 1) models and 2) transport properties. After this, abrief survey of the develop ment of microstructurally-based models is presented. A elose relationship between computer power and speed is suggested. The first group of papers on models covers the subjects of scale and resolution. Most models define and predict characteristics of the pore system, which range in scale from nanometer to millimeter. Various types ofnetworks are proposed in these papers. A good microstructural model must describe the pores and other phases at ascale appropriate to the properties that the model predicts. Also, a good model should be based on fundamental knowledge. In the case of cement-based materials, the important properties may depend on the microstructure, especially the porosity, at several scales.