Ascending or descending locomotion involves a change in potential energy (PE) and a corresponding change in power requirement. We sought to test whether the mechanical power required for steady ascending or descending flight is a simple sum of the power required for level flight and the power necessary for potential energy change. Pigeons (Columba livia) were trained to fly at varying angles of ascent and descent (60 degrees , 30 degrees , 0 degrees , -30 degrees , -60 degrees ), and were recorded using high-speed video. Detailed three-dimensional kinematics were obtained from the recordings, allowing analysis of wing movement. Aerodynamic forces and power requirements were then estimated from kinematic data. As expected, ;PE flight power' increased significantly with angle of flight (0.234 W deg.(-1)), though there appeared to be a limit on the amount of PE that the birds could gain or dissipate per wingbeat. We found that the total power output for flight at various angles was not different from the sum of power required for level flight and the PE rate of change for a given angle, except for the steep -60 degrees descent. The total power for steep descent was higher than this sum because of a higher induced power due to the bird's deceleration and slower flight velocity. Aerodynamic force estimates during mid-downstroke did not differ significantly in magnitude or orientation among flight angles. Pigeons flew fastest during -30 degrees flights (4.9+/-0.1 m s(-1)) and slowest at 60 degrees (2.9+/-0.1 m s(-1)). Although wingbeat frequency ranged from 6.1 to 9.6 Hz across trials, the variation was not significant across flight angles. Stroke plane angle was more horizontal, and the wing more protracted, for both +60 degrees and -60 degrees flights, compared with other flight path angles.