This paper leverages variation in the access to the Mexican railroad network in the early 1900s to estimate its impact on migration to the United States and evaluate its long-run persistence after passenger rail service became obsolete. Using an IV strategy based on least-cost paths between historical cities, we find that locations with railroad access had migration rates four times higher than those without in the early twentieth century. Sequential migration was the key mechanism: railroads first facilitated internal mobility toward railroad hubs, then onward migration to the US. Railroad access also contributed to structural transformation, raising urbanization and local economic development. In terms of persistence, locations with historical railroad access show weakly lower total migration rates to the US in the early 21st century, consistent with local economic growth reducing the incentive to migrate. Yet destination-specific patterns prove remarkably durable: locations that disproportionately sent migrants to California, Arizona, or Texas in the 1900s continued to do so in the 2000s, reflecting the persistence of migrant networks.