Widespread decline of Neotropical migrant songbirds requires better understanding of the mechanisms driving juvenile mortality. We used mark—resight encounter histories of 3,990 banded fledglings (1998–2002) to test whether late breeding or large brood size negatively affected apparent annual and migration—wintering survival probability of juvenile Purple Martins (Progne subis). We estimated apparent annual juvenile survival (φ;_a, fledging to 1 year old) by resighting individuals as adults at regional breeding colonies and at a premigratory roost. We tested for carryover effects of fledge week and brood size on migration—wintering survival (φ;_m, premigratory roost to 1 year old) using two encounter occasions per season (premigratory roost, breeding colony) to partition annual survival into premigration versus migration—wintering survival. Annual survival (± SE) was 0.27 ± 0.027 and the best model (model weight = 0.93) included week of fledging and brood size, with survival probability decreasing with increasing fledge date and brood size. Apparent fledgling survival probability to premigratory roosts (φ;_r) was 0.87 ± 0.03 and migration—wintering juvenile survival (φ;_m) averaged 0.32 ± 0.04 (range: 0.25–0.46 among years). The best model included an effect of fledging week (model weight = 0.99) on migration—wintering survival but little support for brood size effects. Late-fledged young that survive to begin migration may incur higher subsequent mortality because of less foraging time and experience before migration. Telemetry of 15 fledglings in 2007 also revealed high premigration survival to the premigratory roost (0.73). Most (81%) juvenile mortality occurred after the onset of migration. Juvenile recruitment and population dynamics are likely closely linked to migration and wintering ground threats.