This research examines the current processes of land rights registration in Ethiopia, Ghana and Mozambique and assesses their outcomes for vulnerable groups. It shows that land registration is not inherently anti-poor. The distributional consequences of land registration depend on the design of the registration process and of the institutions responsible for its management. It is important to design land registration systems that secure the land rights of marginalised groups in specific geographic and historical contexts, rather than adopting blueprint solutions based on Western models.