Joint moments help us understand joint loading and muscle function during movement. However, the interpretation depends on the choice of reference frame, but the different reference frames have not been compared in dynamic, high-impact sporting movements. We have compared the magnitude and the resulting ranking of hip and knee joint moments expressed in the laboratory coordinate system, the local system of the distal segment and projected or decomposed to the Joint Coordinate System (JCS) axes. Hip and knee joint moments of drop jumps and sidestep cutting in 70 elite female handball players were calculated based on recordings from an eight-camera 240 Hz system and two force platforms and expressed with the four methods. The greatest variations in magnitude between conditions were seen for drop jump hip internal rotation (range: 0.31–0.71 Nm/kg) and sidestep cutting knee flexion (2.87–3.39 Nm/kg) and hip internal rotation (0.87–2.36 Nm/kg) and knee internal rotation (0.10–0.40 Nm/kg) moments. The rank correlations were highest between conditions for flexion moments (0.88–1.00) and sidestep cutting abduction moments (0.71–0.98). The rank correlations ranged from 0.64 to 0.73 for drop jump knee abduction moments and between −0.17 and 0.67 for hip and knee internal rotation moments. Expression of joint moments in different reference systems affects the magnitude and ranking of athletes. This lack of consistency may complicate the comparison and combination of results. Projection to the JCS is the only method where joint moments correspond to muscle and ligament loading. More widespread adoption of this convention could facilitate comparison of studies and ease the interpretation of results.