A number of primary care research networks were set up throughout England in 1998 in order to (1) improve the quality of primary care research (2) increase the research capacity of primary care, and (3) change the culture of primary care towards reflective inquiring practice (NHSE, 2000b). It is not clear how best to operate a network to achieve these diverse aims. This paper describes the first 30 months of a network that adopted a whole system approach in the belief that this would offer the best chance of simultaneously achieving the three aims. A cycle of activity was designed to facilitate the formation of multidisciplinary coalitions of interest for research with complementary 'top down' and 'bottom up' programmes of work co-existing. At least 330 people participated in the generation of research questions of whom one third (33%) were general practitioners, 16% community nurses, 6% practice managers and other primary care practitioners. Over two fifths (43%) were 'key allies'--academics, health authority staff, community workers and project workers. One fifth (110) of all practices (500) in the WeLReN area have collaborated in at least one research project. The ratio of doctor:nurse participation in the 24 research project teams was markedly different in the supported coalitions (2:1) compared to projects devised and led by more experienced researchers (6:1). The evidence suggests that it is possible to operate a primary care research network in a way that develops coalitions of interest from different parts of the health care system as well as both 'top down' and 'bottom up' led projects. It is too early to tell if the approach will be able to achieve its aims in the long-term but the activity data are encouraging. There is a need for more research on the theoretical basis of network operation.