This study aimed to compare mechanical variables derived from torque-cadence and power-cadence profiles established from different cycle ergometer modes (isoinertial and isokinetic) and modelling procedures (second- and third-order polynomials), whilst employing a novel method to validate the theoretical maximal power output (Pmax). Nineteen well-trained cyclists (n = 12 males) completed two experimental sessions comprising six, 6-s maximal isoinertial or isokinetic cycling sprints. Maximal pedal strokes were extracted to construct power-cadence relationships using second- and third-order polynomials. A 6-s sprint at the optimal cadence (Fopt) or optimal resistance (Topt) was performed to assess construct validity of Pmax. No differences were found in the mechanical parameters when derived from isokinetic (Pmax = 1311 ± 415, Fopt = 118 ± 12) or isoinertial modes (Pmax = 1320 ± 421, Fopt = 116 ± 19). However, R2 improved (P < 0.02) when derived from isoinertial sprints. Third-order polynomial modelling improved goodness of fit values (Standard Error, adjusted R2), but derived similar mechanical parameters. Finally, peak power output during the optimised sprint did not significantly differ from the theoretical Pmax in both cycling modes, thus providing construct validity. The most accurate P-C profile can be derived from isoinertial cycling sprints, modelled using third-order polynomial equations.