Critical power (CP) demarcates the boundary between heavy and very heavy exercise intensity domains, and therefore, the power output (PO) that can be sustained at the maximal metabolic steady-state during constant-PO exercise (i.e., maximal lactate steady-state (MLSS)). However, the estimated CP does not always reflect a sustainable intensity of exercise, where blood lactate concentration ([La]) and oxygen uptake (O2) reach a plateau. Objectives: To test cyclists’ ability to predict their highest PO associated with metabolic steady-state based on their own perception of effort. Design: Repeated measures. Methods: Thirteen healthy young cyclists (26 ± 3years; 69.0 ± 9.2 kg; 174 ± 10 cm) were tested. Five time-to-exhaustion trials were used to derive CP based on a 2-parameter hyperbolic model (CPHYP). Participants performed two 30-min rides at a self-selected PO that they considered their highest sustainable exercise intensity (CPSELF). Additionally, MLSS was determined as the highest PO at which variation in [La] ≤1.0 mmol L−1 between the 10th and 30th min was observed during a 30-min ride. Results: Mean PO at CPSELF (233 ± 42 W) was similar (p > 0.05) to MLSS (233 ± 41 W), whereas CPHYP (253 ± 44 W) consistently overestimated (p < 0.05) the PO associated to metabolic steady-state. The limits of agreement (LOA) between MLSS and CPSELF were −20 to +20 W (bias = 0 W, p > 0.05), whereas the LOA between CPHYP and CPSELF were −40 to 0 W (bias = −20 W, p < 0.05). CPSELF and MLSS presented similar (p > 0.05) metabolic response (i.e., O2, [La], and HR). Conclusions: Compared to CPHYP, CPSELF may offer a more precise approach to predict the constant-PO associated with maximal physiological steady-state.