Survival, distribution, and behavior of hatchery (n = 493) and naturally reared (n = 133) smolts of Atlantic salmon Salmo salar migrating through the Penobscot River and estuary in Maine were evaluated with acoustic telemetry in 2005 and 2006. Survival and use of a secondary migration path (the Stillwater Branch) were estimated with a multistate mark–recapture model. Higher rates of mortality per kilometer (range = 0.01–0.22) were observed near release sites and within reaches that contained three particular dams: Howland, West Enfield, and Milford dams. Estimated total survival of tagged hatchery smolts through entire individual reaches containing those dams ranged from 0.52 ( 0.18) to 0.94 ( 0.09), whereas survival through most of the reaches without dams exceeded 0.95. Of those smolts that survived to the Penobscot River–Stillwater Branch split at Marsh Island, most (≥74%) remained in the main stem around Marsh Island, where they experienced lower survival than fish that used the Stillwater Branch. Movement rates of hatchery-reared smolts were significantly lower through reaches containing dams than through reaches that lacked dams. Smolts arriving at dams during the day experienced longer delays than smolts arriving at night. Planned removal of two dams in this system is expected to enhance the passage of smolts through the main-stem corridor. However, the dams currently scheduled for removal (Great Works and Veazie dams) had less influence on smolt survival than some of the dams that will remain. This case study shows that by examining prerestoration migration dynamics throughout entire river systems rather than just in the vicinity of particular dams, tracking studies can help prioritize restoration efforts or predict the costs and benefits of future hydrosystem changes.