Little is known about how in vivo muscle efficiency, that is the ratio of mechanical and metabolic power, is affected by changes in locomotory tasks. One of the main problems with determining in vivo muscle efficiency is the large number of muscles generally used to produce mechanical power. Animal flight provides a unique model for determining muscle efficiency because only one muscle, the pectoralis muscle, produces nearly all of the mechanical power required for flight. In order to estimate in vivo flight muscle efficiency, we measured the metabolic cost of flight across a range of flight speeds (6–13 m s−1) using masked respirometry in the cockatiel ( Nymphicus hollandicus ) and compared it with measurements of mechanical power determined in the same wind tunnel. Similar to measurements of the mechanical power–speed relationship, the metabolic power–speed relationship had a U-shape, with a minimum at 10 m s−1. Although the mechanical and metabolic power–speed relationships had similar minimum power speeds, the metabolic power requirements are not a simple multiple of the mechanical power requirements across a range of flight speeds. The pectoralis muscle efficiency (estimated from mechanical and metabolic power, basal metabolism and an assumed value for the ‘postural costs’ of flight) increased with flight speed and ranged from 6.9% to 11.2%. However, it is probable that previous estimates of the postural costs of flight have been too low and that the pectoralis muscle efficiency is higher.