Primary migration is a complex process involving coupled fluid generation and mass transfer in low permeable rocks. Joint experimental and modelling work have been performed in order to better understand this problem. Eventually, experimental results have been modelled and extrapolated to geological conditions. Maturation and expulsion experiments have been carried out with water saturated immature Type II source rocks from the Paris Basin, with sample weights up to 3 kg. Temperatures were stepwise increased from 20°C to 285, 330 or 380°C at constant hydrostatic pressures of respectively 17 or 55 MPa during up to 5000 h. During the experiments, generated fluids expelled out of the heating cell and were sampled prior to a complete analysis. Mass balances were checked on hydrocarbon gases, liquid hydrocarbons, CO 2 and water. Meanwhile, closed system pyrolysis experiments were performed on isolated kerogen extracted from the same source-rocks. They allowed us to derive a compositional kinetic cracking reaction network. It was then possible to tackle the expulsion of fluids out of the source rock: the kinetic scheme provides a guideline to understand the generation of fluids from kerogen; the expulsion experiments provide an insight into the expulsion mechanisms. No axial stress has been applied to the samples in the experiments. Therefore, the influence of rock compaction, due to effective stress, was not studied in these experiments. A numerical model of expulsion has been set up based on volume and mass balance conservation equations in the source rock. The volume is assumed to vary as solid kerogen transforms to liquid products and solid residue. The mass of each compound is kept in balance, as it is created or destroyed through primary and secondary cracking and expelled through multiphase fluid flow. Therefore, depending on pressure, temperature and composition, fluids in the source rock are split into several phases, following a thermodynamical equation of state; all mobile phases flow according to Darcy's law. Experiments were modelled, the best fit being achieved with the following conditions: • • all fluids, including water, CO 2 and hydrocarbons are allowed to form a multiphase mixture; • • the heavy compounds (resins and asphaltenes) are in a single very viscous phase and are less easily expelled, as most of them are cracked inside the source rock. Extrapolations to geological pressure and temperature conditions were performed afterwards, in order to check the phase behaviour and the expulsion efficiency.