To examine the feasibility of using near-infrared spectroscopy (NIRS) for physiological threshold detection and whether NIRS-derived parameters differ between highly-trained and less-trained cyclists. Twenty-seven male cyclists were divided into: highly trained endurance cyclists (EA) and recreational cyclists (RA). Participants performed a step-incremental cycling test to exhaustion. Ventilatory thresholds (VT1 and VT2) were determined using gas-exchange variables. NIRS sensor was placed on the vastus lateralis muscle to identify breakpoints corresponding to ventilatory thresholds. No significant differences were observed between NIRS-derived thresholds, compared to VT1 and VT2 (F = 1.04–1.33, p = 0.26–0.36). Moderate to strong correlations were found between NIRS-derived thresholds and ventilatory thresholds (r = 0.65–0.9, p textless 0.01). A moderate correlation was found between maximal oxygen uptake and minimal tissue saturation index (TSI) value during the test (r = − 0.411, p = 0.037). EA group showed tendency towards lower minimal TSI values compared to RA group (MD = 5.46% TSI, p = 0.081). NIRS is a feasible tool for non-invasive assessment of ventilatory thresholds during incremental exercise. TSI, in particular, showed lower variability compared to other NIRS-derived parameters, and may therefore be more suitable for practical applications in sport science. Highly trained athletes demonstrated distinct physiological responses compared to recreational athletes, suggesting enhanced peripheral oxygen extraction.