Video recordings (60 Hz) of horses jumping a 4.5 m wide water jump were made during the team show jumping competition at the 1992 Olympic Games. The tapes were digitised and the results were analysed using the Ariel Performance Analysis System software and Statistical Packages for the Social Sciences. Eighteen horses were studied: 12 that cleared the water jump (Group 1) and 6 that failed to clear the width of the jump (Group 2). Four variables that influence the horizontal distance jumped were measured: horizontal velocity, vertical velocity, height of the centre of gravity (CG) at lift off and height of the CG at landing. The angle of projection was computed from the horizontal and vertical velocities. The angle of projection (Group 1: 15.2° ± 2.2°, Group 2: 12.2° ± 1.8°, F=0.010) and the vertical velocity of the CG (Group 1: 2.03 ± 0.22 m/s, Group 2: 1.56 ± 0.24 m/s, F=0.001) were significantly different between the 2 groups. There were no differences in the height of the CG at lift off (Group 1: 1.36 ± 0.08 m, Group 2: 1.40 ± 0.05 m), the height of the CG at landing (Group 1: 1.35 ± 0.12 m, Group 2: 1.30 ± 0.11 m) or the horizontal velocity of the CG (Group 1: 7.59 ± 0.60 m/s, Group 2: 7.48 ± 0.11 m/s). It was concluded that, to be a successful water jumper, the horse must generate a large vertical velocity during the take off.