Internal mechanical work during cycling, required to raise and lower the legs and change their velocities, is shown to be an important factor when interpreting physiological responses to cycle ergometer exercise. The internal work required to move the legs during concentric and eccentric cycle ergometry at different speeds and workloads was calculated from segmental energy changes determined using cinematography and directly using an eccentric ergometer. The mean internal work rates obtained at pedal frequencies of 30, 60 and 90 min-1 were 11.5, 20 and 62 W respectively. When these estimates were added to the external work rates, they increased concentric and decreased eccentric work rates. The largest differences were seen at low work rates and high pedal frequencies during which concentric work rates increased by 51% and eccentric decreased 60% by the inclusion of internal work. When comparisons of concentric and eccentric cycling at equal uncorrected work rates were made, neglecting to include internal work introduced errors ranging from 12 to 97%. The calculated estimates of internal work agreed well with the power supplied by the eccentric ergometer to move the legs passively. The investigations show that the inclusion of internal work is important when comparing physiological responses during concentric and eccentric ergometry, especially when pedal frequencies exceed 60 min-1 and when work rates are small.