There is a paucity of papers dealing with the system characteristics of laboratories or, put in another way, the institutional character of research organisations. Neither R & D economics nor the sociology of science, as traditionally conceived, has made much headway in providing insight into sets of R & D laboratories and their evolution. Drawing upon an empirical study in the plant breeding and biochemical industry, this paper presents a typology of public research laboratories which is based on three dimensions: scientific production and visibility, type of funding (public or private) and homogeneity of research themes. Three types of public laboratory emerge: the first, called “research centres for the profession”, is composed essentially of laboratories with close ties with small and medium firms (SMEs) and industry associations. The second, called “designers of generic tools and methods”, is oriented towards basic research and themes of general interest to the industry as a whole. The third type, called “basic and specialised laboratories”, strives to develop its scientific visibility. Contracts between this type of laboratory and industry are mainly bilateral and demonstrate the complementarity between public and private research. Each type of laboratory develops specific types of relationship with private partners. The authors have identified three logics underlying these relationships: proximity, market and club. The main objective of contracts based on a proximity logic is to test a hypothesis, while the knowledge produced is mostly tacit and specific. By contrast, knowledge is entirely coded and specific in the market logic, where the aim of the contract is to implement expertise in order to relieve a scientific bottleneck. In a club logic, the aim of contracts is to produce a technical referent. In each kind of contract, the learning trajectories, modes of co-ordination, role of trust and degree to which contracts are complete or not, are different.