During high-intensity running, the oxygen uptake (V over dot O-2) kinetics is characterised by a slow component which delays the attainment of the steady-state beyond the 3rd min of exercise. To assess if the aerobic energy cost of running measured at the 3rd min (C-3) adequately reflects the variability of the true aerobic energy cost measured during the steady-state (C-ss), 13 highly-trained runners completed sessions of square-wave running at intensities above 80% maximal oxygen uptake ((V) over dot O-2 max) on a level treadmill. To evaluate the time at which the steady-state (V) over dot O-2 was attained (t(ss)), the (V) over dot O-2 responses were described using a general double-exponential equation and t(ss) was defined as the time at which (V) over dot O-2 was less than 1% below the asymptotic value given by the model. All the subjects achieved a steady state for intensities equal to or greater than 92% (V) over dot (2 max) and 8 out of 13 achieved it at 99% (V) over dot O-2 max In all cases, t(ss) was less than 13 min. For intensities greater than 85% (V) over dot O-2 max, C-ss was significantly higher than C-3 and was positively related to % (V) over dot O-2 max (r = 0.44; P < 0.001) while C-3 remained constant. The C-3 only explained moderately the variability of C-ss (0.39 < r(2) < 0.72, depending on the velocity or the (relative intensity at which the relationship was calculated). Moreover, the excess aerobic energy cost of running the (difference between C-ss and C-3) was well predicted by age (0.90 < IL < 0.93). Therefore, when the aerobic profile of runners is evaluated, it is recommended that their running efficiencies at velocities which reflect their race intensities should be determined, with (V) over dot O-2 data being measured at the true steady-state.