Purpose: The theoretical velocity associated with (V) over dot O-2max (v(V) over dot O-2max) defined by Daniels (1985) is extrapolated from the submaximal (V) over dot O-2-velocity relationship. (V) over dot O-2 is generally determined by assuming that the aerobic response reacts like a linear first-order system at the beginning of square-wave exercise, with a steady-state reached by the 4th minute. However, at supra-ventilatory threshold work rates, the steady state in (V) over dot O-2 is delayed or not attained. Methods: The present study was carried out to compare three values for v(V) over dot O-2max, determined with Daniels' method, but with (V) over dot O-2 either measured at the 4th minute (v(V) over dot O-2max4), the 6th minute (v(V) over dot O-2max6), or after the attainment of the true steady-state (v(V) over dot (2maxSS)). The metabolic response during square-wave exercise at each of the three v(V) over dot O-2max were also assessed. Results: These velocities were significantly different (P < 0.05), but v(V) over dot O-2maxSS and v(V) over dot O-2max6 were highly correlated (r =0.98; P < 0.05). Blood lactate concentrations measured after exercise at velocities very close to the three v(V) over dot O-2max were similar and the end-exercise (V) over dot O-2 were not different from (V) over dot (2max), but the time required to elicit 95% (V) over dot O-2max during these three square-wave rests were significantly different. Conclusion: Therefore, when v(V) over dot O-2max is determined by extrapolation from the submaximal (V) over dot O-2-velocity relationships, submaximal (V) over dot O-2 should be measured beyond the 6th minute of square-wave exercise (at least if it takes 30 s to reach the desired velocity) to ensure that all v(V) over dot O-2max reported in future studies describe a similar quantitative index.