A paradox of avian long-distance migrations is that birds must greatly increase their body mass prior to departure, yet this is presumed to substantially increase their energy cost of flight, particularly given their tendency to fly relatively fast. However, our understanding of the interactive effects of speed and weight is limited on flight energy costs are based on a few constrained wind tunnel studies. The literature is missing experiments on free-ranging birds flying at similar speeds but with different body masses to test whether the additional energy costs of flying when heavier are very small as indicated by laboratory experiments, or are hyper-allometric as might be traditionally expected. We addressed this by undertaking a novel design of loading experiment on free-flying homing pigeons which utilised thin ventrally located weight and a heart rate data recorder. Our data show that when homing pigeons flying in a flock are loaded with ventral weight, both their heart rate and estimated energy expenditure barely rise. The net effect is that energy costs per unit time increase only slightly and per unit mass they decrease. We suggest that this is because homing pigeon flights are relatively fast, and consequently flight costs associated with increases in body parasite drag dominate over those of weight support, leading to an improvement in mass-specific flight economy. We propose that the small absolute aerodynamic penalty for carrying enlarged fuel stores and flight muscles during fast flight has helped to select for the evolution of long-distance migration.