Inspired by Ostrom’s concept of polycentric governance, this article aims to refine the analytical framework through which contemporary access to land is analysed. By drawing on extensive fieldwork and conducting a review of the existing literature on the making and implementation of Tanzania’s 1999 land reform, it challenges some of the main assumptions behind the land access and land grabbing literatures about the level at which agency is placed. Processes governing access to land are more contingent than they are most often depicted, involving actors at the local, national and international levels. National and local level actors are often more important than, in particular, the land grab literature tends to suggest. This implies that the state should be seen not merely as a site of ‘legitimate theft’, but also as one in which rights may be upheld. Based on the experience of Tanzania, the article suggests that analytical a priori assumptions about where agency is placed should be abandoned and replaced with empirical research into the relations between actors at all levels.