The standard theory of property rights views the establishment of individual, private property rights as a major condition for a secure and optimal allocation of the productive resources (Demsetz, 1967; Johnson, 1972; Alchian and Demsetz, 1973). In numerous developing countries, this approach led to policies aiming at substituting land regulations through "formal" institutions (positive law, property rights formalized by a title, enforcement through the judicial system) to land regulations organized through "informal" institutions (« customary law», very « imperfect » property rights, enforcement through traditional institutions). However, the implementation of land titling programs made clear the limitations of such policies with respect to the recognition of land rights, especially in African contexts. Today, development experts mostly agree that land tenure security can be achieved in a more effective, simpler and cheaper way by involving local communities rather than by a direct and heavy State intervention through a titling program (Atwood, 1990; Stanfield 1990; Platteau, 1996; Bruce and al., 1994; Deininger & Binswanger, 2001; Banque Mondiale, 2003; Van Den Brink & al., 2006; Deininger & Feder, 2009). The policy of decentralized land certification implemented in Madagascar since 2004 illustrates this new paradigm. Municipal "guichets fonciers" deliver land certificates after a quick and simple procedure of recognition of the rights and right-holders. The key issue tackled in this paper is the actors’ strategies regarding land certification as a way to secure land rights, in a context of legal and institutional pluralism. We focus on land certificates as an alternative to other securization devices: do land certificates supplement or substitute other devices such as customary regulations, « petits papiers », or titling? Our analysis relies on first-hand data collected in 2008 and in 2010 in 7 villages located in the Commune rurale of Faratsiho, in the Central highlands, combining qualitative (participatory observations, in-depth interviews of 50 households) and quantitative (questionnaire survey of 400 households and 2500 plots) methods.