A fundamental prerequisite for designing public policy for farmers is that we know who they are. In the case of India, despite agriculture’s continuing importance as a source of livelihood, there is little understanding of how many farmers there are in the country and who they are. This paper suggests that there are three constructs of farmers – a statistical construct – that national agencies deploy for different surveys, a programmatic construct – that governments use to define beneficiaries and a construct that derives from the complex reality of what farming entails, that researchers often rely on. In the context of India these constructs don’t often align with one another. This has important implications for policies that end up, often by construction, being exclusionary. This paper suggests the need for a coherent set of definitions. Too often, researchers call for collection of more or better data in response to the limitations of existing surveys. We argue that existing surveys can be leveraged to generate better estimates that match not just the requirements of programmatic goals but also to reflect the realities of farmer identities better.