Optimal migration theory suggests specific scaling relationships between body size and migration speed for individual birds based on the minimization of time, energy, and risk. Here we test if the quantitative predictions originating from this theory can be detected when migration decisions are integrated across individuals. We estimated population-level migration trajectories and daily migration speeds for the combined period 2007–2011 using the eBird data set. We considered 102 North American bird species that use flapping or powered flight during migration. Many species, especially in eastern North America, had looped migration trajectories that traced a clockwise path with an eastward shift during autumn migration. Population-level migration speeds decelerated rapidly going into the breeding season, and accelerated more slowly during the transition to autumn migration. In accordance with time minimization predictions, spring migration speeds were faster than autumn migration speeds. In agreement with optimality predictions, migration speeds of powered flyers scaled negatively with body mass similarly during spring and autumn migration. Powered fliers with longer migration journeys also had faster migration speeds, a relationship that was more pronounced during spring migration. Our findings indicate that powered fliers employed a migration strategy that, when examined at the population level, was in compliance with optimality predictions. These results suggest that the integration of migration decisions across individuals does result in population-level patterns that agree with theoretical expectations developed at the individual level, indicating a role for optimal migration theory in describing the mechanisms underlying broadscale patterns of avian migration for species that use powered flight.