Human migration is often considered an important driver of land use change and a threat to protected area integrity, but the reasons for in‐migration, the effectiveness of conservation restrictions at stemming migration, and the extent to which migrants disproportionately contribute to land use change has been poorly studied, especially at fine spatial scales. Using a case study in eastern Madagascar (603 household surveys, mapping agricultural land for a subset of 167 households, and 49 focus group discussions and key informant interviews), we explore the patterns and drivers of migration within the lifetime of those currently alive. We investigate how this influences forest conversion on the border of established protected areas and sites without a history of conservation restrictions. We show that in‐migration is driven, especially in sites with high migration, by access to land. There is a much higher proportion of migrant households at sites without a long history of conservation restrictions than around long‐established protected areas, and migrants tend to be more educated and live closer to the forest edge than non‐migrants. Our evidence supports the engulfment model (an active forest frontier later becoming a protected area); there is no evidence that protected areas have attracted migrants. Where there is a perceived open forest frontier, people move to the forest but these migrants are no more likely than local people to clear land (i.e., migrants are not “exceptional resource degraders”). In some parts of the tropics, out‐migration from rural areas is resulting in forest regrowth; such a forest transition is unlikely to occur in Madagascar for some time. Those seeking to manage protected areas at the forest frontier will therefore need to prevent further colonisation; supporting tenure security for existing residents is likely to be an important step.