Critical power (CP) is an important training threshold and represents the highest power output that elicits steady-state physiological responses. Research suggests that CP and the finite work capacity (W'), can be estimated from a single 3-min bout of all-out cycling. Five experimental studies were undertaken to explore the reliability and validity of CP tests, and to propose a novel all-out testing protocol. Study one investigated the reliability and validity of the 3-min cycling test when performed against a fixed resistance and in isokinetic mode. Results suggested that the 3-min cycling test provided a reliable and valid estimate of CP in isokinetic mode, but significantly overestimated CP when performed against a fixed resistance. Study two investigated the effect of cadence on CP and W' during the 3-min cycling test when performed against a fixed resistance, with results suggesting that a better estimation of CP is observed at higher cadences (e.g. preferred cadence +10 rev·min-1). Studies three and four focused on measuring power output using cycle-mounted power meters to support the novel all-out testing protocol used in study five. The PowerTap P1 pedals demonstrated greater reliability and validity than the Garmin Vector 2 pedals across all power outputs, with reliability maintained after prolonged use. Consequently, the PowerTap P1 pedals were used in study five, which investigated the reliability and validity of a novel all-out cycling test to estimate CP and W'. Results suggested that CP could be estimated from the novel all-out cycling test; however, caution should be taken when estimating W'. The results also suggested that cycling at CP calculated from the original protocol, 3-min cycling test protocol, and novel all-out test protocol resulted in exhaustion occurring within 20 min, and a metabolic steady-state was not observed. The overall findings of this thesis question the underpinning physiology of CP, and whether CP represents the boundary between the heavy and severe exercise intensity domains.