Transformational leadership is associated with positive athlete outcomes, but existing are criticized for focusing on outcomes rather than behaviors and lacking sport specificity. This study aimed to develop an instrument for measuring transformational leadership in sport using observation studies and coach-specific behaviours to address these limitations. Phase 1 involved literature review and item generation, resulting in 51 items that described relevant coach behaviors and aligned with existing observation instruments. In Phase 2, initial item validation, items were rated by a group of 24 sport students in terms of specificity, clarity, and relevance. Subsequently, interviews were conducted with 15 students and two expert coaches about how they interpreted each item before two academic experts reviewed the final items to ensure they matched the criteria. In Phase 3, 233 Norwegian youth team sport athletes aged 13-19 were sampled for the purpose of item-reduction (to 16 items) and identification of subdimensions. EFA’s (MLR estimator, Geomin rotation) were performed for the 29 items. The four-factor solution (reflecting the four I's in transformational leadership) yielded the best model-fit, and the four highest loading items were retained for each factor in a stepwise process starting with the factor with the highest eigenvalue. The 16-items EFA provided good model fit (S-B χ2 = 71.863 [df = 62, N = 233], p = .184; CFI = .995; TLI = .991; RMSEA = .026 [.000 – .049], and SRMR = .015). Further data collection is in progress, and a subsequent independent sample is needed to confirm the suggested structure.