The article focuses on the generation of knowledge about social relationships within participatory rural development projects. In the Kribhco Indo-British Rainfed Farming Project (KRIBP) in India significant constraints were experienced in using PRA techniques for social analysis. Although they were successful at generating agro-ecological information, they were less helpful in revealing the structures of power and influence within a village or in helping project workers identify the social relations which shaped 'public' opinion. It is unlikely that public participatory research methods will prove good instruments for the analysis of local power relations since they are shaped by the very social relations which are being investigated. In fact, an understanding of local social networks is a necessary prerequisite for the organisation of effective PRA based work. In KRIBP understanding of these relations came from participant observation and critical review of the PRA activity itself rather than from the use of PRA methods.