This study characterized the profile of near-infrared spectroscopy (NIRS)-derived muscle deoxygenation (Δ[HHb]) and the tissue oxygenation index (TOI) as a function of absolute (PO(ABS)) and normalized power output (%PO) or oxygen consumption (%VO(2)) during incremental cycling exercise. Eight men (24 ± 5 year) each performed two fatigue-limited ramp incremental cycling tests (20 W min(-1)), during which pulmonary VO(2), Δ[HHb] and TOI were measured continuously. Responses from the two tests were averaged and the TOI (%) and normalized Δ[HHb] (%Δ[HHb]) were plotted against %VO(2), %PO and PO(ABS). The overall responses were modelled using a sigmoid regression (y = f ( 0 ) + A/(1 + e(-(-c+dx)))) and piecewise 'double-linear' function of the predominant adjustment of %Δ[HHb] or TOI observed throughout the middle portion of exercise and the 'plateau' that followed. In ~85% of cases, the corrected Akaike Information Criterion (AIC(C)) was smaller (suggesting one model favoured) for the 'double-linear' compared with the sigmoid regression for both %Δ[HHb] and TOI. Furthermore, the f ( 0 ) and A estimates from the sigmoid regressions of %Δ[HHb] yielded unrealistically large projected peak (f ( 0 ) + A) values (%VO(2p) 114.3 ± 17.5; %PO 113.3 ± 9.5; PO(ABS) 113.5 ± 9.8), suggesting that the sigmoid model does not accurately describe the underlying physiological responses in all subjects and thus may not be appropriate for comparative purposes. Alternatively, the present study proposes that the profile of %Δ[HHb] and TOI during ramp incremental exercise may be more accurately described as consisting of three distinct phases in which there is little adjustment early in the ramp, the predominant increase in %Δ[HHb] (decrease in TOI) is approximately linear and an approximately linear 'plateau' follows.