Farmer participatory research (FPR) methods have been advocated as a means of increasing the client focus of agricultural research in developing countries. The National Agricultural Research Systems (NARS) in these countries have adopted them to varying degrees--often as an implicit conditionality of donor supported research projects. This paper seeks to demonstrate that, despite the apparent acceptance of FPR in NARS, the fundamental nature of the relationship between scientists and farmers remains unchanged. FPR has largely failed in its attempts to improve the efficiency of agricultural research by restructuring science/production relations. This failure is the result of the 'systems problem' in agricultural research, whereby the complex interrelationship of actors, institutions and resources prevents FPR methods being compatible with NARS.