This paper first defines the three functions of a national research system: orientation, programming, and research. It then defines the two main possible configurations for such research systems according to their degree of integration vs. separation of the functions, in particular concerning the programming and research functions (distinguishing between cases with an important role for public research organizations and those with an important role for funding agencies). It reveals differences among countries in this respect. The case of France is described, with its integrated model in place since the end of WW2. The paper shows the success of this model. However, it also discusses issues that have been raised since the 1990s, including the growing role of the universities, the dynamics of research toward interdisciplinarity, the need for transparency, and the emergence of the European Research Space, none of which is easily addressed by the integrated model. It describes the 2005-2008 reforms that made the system evolve toward one in which functions are separated. In conclusion, the paper offers possible scenarios for relationships between universities and public research organizations. JEL codes: I23, I28, 038