Departure decisions of songbirds at ecological barriers they encounter en route can strongly influence time, energy and survival costs of migration. To date, most field studies of departure decisions and their correlates have used indirect methods and followed migrants at a single stopover site, with limited information on what happens to individuals after they depart from the site. We used an automated radiotelemetry array extending 350 km from southwest Nova Scotia to southern Maine to study the migratory and stopover movements of Northern Waterthrushes Parkesia noveboracensis, Red‐eyed Vireos Vireo olivaceus and Yellow‐rumped Warblers Setophaga coronata in relation to fuel load and weather at the northeastern edge of the Gulf of Maine. From the 105 radio‐transmitters we deployed in southwest Nova Scotia, we recorded 42 landscape‐scale stopover flights and 47 migratory flights by 75 individuals. Of the migratory flights, 57% were orientated southwest, a trajectory that, if held, would require individuals to complete a 350–440 km overwater flight. The remaining 43% of migratory flights were orientated northwest, away from the Gulf of Maine, and 15 individuals were confirmed to have detoured around all or a portion of the barrier, as evidenced by their being re‐detected over the Bay of Fundy and/or along the coast of Maine between 4 h and 15 days later. Across all individuals, initial fat score had a positive effect on departure probability, especially for individuals that made stopover flights. Among weather variables, tailwind assistance was the best predictor of migratory departures but did not appear to be the main factor determining whether individuals orientated towards or away from the Gulf of Maine. Weather had little effect on departure decisions of individuals that made stopover flights. These differences in the correlates of migratory departures and stopover flights would probably not have been distinguishable had our study been restricted to a local scale. Therefore, our findings highlight the importance of expanding the scale at which departure decisions and the ecology of stopover in general are studied.