PURPOSE:This study aimed to assess the validity of the first (EMGth1) and second (EMGth2) breakpoints in the EMG signal during skating. SCOPE:Ten well-trained long track speed skaters performed a maximal incremental skating test on a slide board. EMG signals from six lower limb muscles were recorded during the last 15 s of each stage and converted to Root Mean Square for determination EMGth1 and EMGth2 using mathematical (2 and 3 linear regressions) and visual methods. CONCLUSIONS:EMGth1 had a low detection rate (<50%) while EMGth2 could be identified visually in > 80% of cases, in 85% of cases using 2-lines and 63% using 3-lines regression. Quads (VL + VM) and Gluts (GM + GMd) had the highest EMGth2 detection rate for all methods (>70%). EMGth2 from Quads and Gluts detected by the 2-lines and 3-lines regression were not different than the second ventilatory threshold (VT2) (p > 0.05), while the visual method overestimated VT2 (p < 0.01). EMGth2 detected from Quads by the 2-lines regression method presented better correlation with VT2 stage (r = 0.91), lowest bias, and limit of agreement. We conclude that EMG is a valid non-invasive method to detect VT2 during skating when using a mathematical method to determine EMGth2.