This study examined the association between riding speed and elimination in Fédération Equestre Internationale (FEI) endurance events. A total of 35,061 horse starts from 1st July 2012 to 31st December 2015 were included in a multivariable logistic regression model containing 25 different risk factors. Riding speeds in individual stages (‘loops’) were included as individual risk factors in studying the progress of horses through loops 1–3 of each endurance ride. The possibility of real-time, ‘mid-ride’ predictive modelling was explored by modelling both riding speeds and sudden changes in speed between loops as potential risk factors. Faster riding speeds, especially during loops 1 and 2, were associated with deleterious outcomes. Furthermore, sudden drops in riding speed during loop 3 were associated with an increased likelihood of elimination.