Background Most Flow measurement tools have mainly demonstrated their usefulness for the scientific study of variations in optimal experience. However, with the exception of the Flow Questionnaire (FlowQ, Csikszentmihalyi, 1975, 1982), very few of them are really able to reveal flow state as a cut-off point. Aims Identify a method closest to concordance with FlowQ, specifically adapted for training contexts. Method -Data collection (n = 1729) using the flow in education scale (EduFlow-2, Heutte et al., 2021), including 4 sub-dimensions (Cognitive control (D1); Immersion & Time Transformation (D2); Loss of self-consciousness (D3); Autotelic experience (D4)), and the 3 items of FlowQ. -3 Structural equations modelling (SEM1, SEM2, SEM3) with a 2nd order factor based on the EduFlow-2 data. -Establishment of 3 formulas (F1, F2, F3) taking into account the weight of the first-order factors on the 2nd-order factor (Flow state (FS)) to establish 3 scores (SF1, SF2, SF3) per individual. -F1: based on SEM1 factors = 4 sub-dimensions of EduFlow-2 connected on FS -F2: based on Kawabata and Mallett’s (2011) model, which distinguishes flow conditions from flow state, and SEM2 factors = only D2, D3 and D4 connected on FS. -F3: based on SEM3 factors = 4 sub-dimensions of EduFlow-2, but only D2, D3 and D4 connected on FS. Results These 3 methods allow flow state detection in concordance with FlowQ, as follows: 68,5% (F1), 71,9% (F2), 71,7% (F3) Discussion Methodological investigations need to be continued, but these first results open up new research prospects, particularly in the field of lifelong learning.