ABSTRACT The aims of this study were to ensure that the skating velocity describes a mono-exponential function in order to determine the reliability of radar-derived profiling results from skating sprint accelerations applying sprint running force-velocity assessment approach. Eleven young highly-trained female ice hockey players performed two 40-m skating sprints on two separate days to evaluate inter-trial and test-retest reliability. The velocity-time data recorded by a radar device was used to calculate the kinetics variables of the skating sprint acceleration: maximal theoretical force (F0), maximal theoretical velocity (V0), maximal theoretical power (Pmax) and the slope of the linear force-velocity relationship (SFV). SFV and SFVrel variables (the slope of the linear relationship between horizontal force relative to body mass and velocity) demonstrated ‘low’ to ‘moderate’ intra-class correlation coefficients (ICC). All other variables revealed ‘acceptable’ inter-trial and test-retest reliability (ICC ≥ 0.75 and coefficient of variation [CV] ≤ 10%). Furthermore, test-retest reliability (ICC and CV) and sensitivity [Standard Error of Measurement (SEMs) ≤ Small Worthwhile Change (SWCs)] were higher when averaging the two trials compared to the best trial (40-m split time) only. These findings offer a promising and simple method to monitor training-induced changes in macroscopic mechanical variables of ice hockey skating performance.