The state of the art in modeling groundwater systems is related to developments in geology, physics, geophysics, hydrogeology, and mathematics. Too often, the research in these areas is pursued independently by different teams and is therefore not integrated around a unique vision. The aim of the ENSEMBLE project is therefore to integrate the recent developments in quantitative geology, hydrology, numerical, and stochastic modeling to increase our ability to understand and predict complex hydrological systems. The teams involved in the ENSEMBLE project (ETH Zürich, Stanford University, Universities of Basel, Bern, Neuchâtel and Lausanne) will focus their research on developing new integrated methods (multiple point statistics, joint hydrogeological and geophysical inversion, fast methods for uncertainty analysis, etc.) for the characterization of the heterogeneity of alluvial systems and for the efficient stochastic ensemble simulation of groundwater flow and solute transport in those systems. Four common field sites corresponding to different alluvial systems (Tagliamento, Thur, Birs, Herten) will be used to illustrate and test the concepts and methods developed in the project.