Great advances were made during the decade of 1975–1985 in the understanding of HC migration by establishing that the main mechanisms were polyphasic flow, whatever its stage, and that migration in solution in water and/or by molecular diffusion was insignificant, except for methane, ethane and benzene. Research is now focusing on points which still have not been satisfactorily answered such as: 1. (1) Behaviour of source-rocks during HC expulsion: the main driving force for fluid expulsion has been recognized to be excess fluid pressure within the source-rock with respect to its surrounding environment of permeable beds. The pressure build-up is due mainly to accommodation by the fluids of part of the mechanical stresses which are exerted on the rock (mainly sedimentary load but also stresses developed by tectonics), to thermal expansion of water, and to HC formation from kerogen in response to increasing temperatures. However, it is not clear whether or not petroleum expulsion proceeds preferentially along oil-wet networks (kerogen or oil-wet mineral surfaces). Nor it is not clear whether or not microfracturing, provoked by an internal fluid pressure higher than allowed by the mechanical strength of source-rocks, is necessary for HC expulsion. 2. (2) Modifications of the compositions of HC phases (e.g. “oil” versus “gas”, or the distribution of biomarkers during migration,) and losses of hydrocarbons along the migration pathways. Although careful observations of natural sample series, and also experimentation, have been done, no clear picture has emerged for the moment. However, compositional effects seem to be due more to changes in thermodynamic conditions along migration pathways, which provoke inter alia hydrocarbon phase separation (retrograde condensation), than to differential retention by mineral surfaces. It is likely that, while careful observations on a large variety of source-rock samples are needed to more clearly understand source-rocks functioning, progress in understanding HC migration will come increasingly from modeling and experimentation. Models, which mathematically simulate natural processes, are necessary to help the human mind in handling such complex phenomena and therefore are of precious help to focus research on the most important points. They replace migration in its geological and historical dimensions with numerical or conceptual tools, while observation only describes the present state. Models are also the tools by which knowledge on migration can be applied to exploration problems. However quantitative applications of models require numerous and precise geological data that are difficult to gather at early stages of exploration. In the field of experimentation, it will be easier to design crucial experiments, now that HC migration mechanisms are better understood. Thanks to experimentation, it should also be possible to validate models. However, so as to reach such aims, substantial progress has to be made in the technology of experimental systems.