Quantification of external training load for sprint kayak athletes can be challenging due to the influence of the water flow on boat velocity in a flowing river environment. Therefore, this study examined the utility of novel measures of power output (PO) and its relationship to measures of relative boat speed when training on a flowing river. Twelve (8 males, 4 female) well-trained sprint kayak athletes completed 4 separate on-water sessions comprising one time-trial session (2 × 1000-m maximal efforts) and three repeated sprint kayak training sessions (5 x split 1000-m [2 × 500-m up and down the river] submaximal efforts) in their individual (K1) kayak. For each session, a Kayak Power Meter recorded athletes’ PO, and a SpeedCoach device recorded relative land-speed via a Global Positioning System (GPS) (S GPS), and relative water-speed via an impeller mounted under the boat hull (S IMP). Non-linear least squares regression were used to evaluate the curvilinear relationship between PO and speed (S GPS and S IMP) data. The exponents of velocity in the PO-S IMP relationship (2.87 females, 2.94 males) were closer to theoretical values (3.00) and showed greater model accuracy (root mean squared error (RMSE) = 20–26 W) than the PO-S GPS relationships (speed exponents = 1.58–2.02, RMSE = 31–40 W). Overall, PO measures could better account for the influence of water flow compared to traditional S GPS measures, and therefore, may be more suitable for quantifying athletes’ external load in their training environment. Highlights Since traditional S GPS and time-to-completion measures do not adjust for the water flow, these measures appear limited for prescribing and monitoring sprint kayak training within flowing river environments. The prescription of paddling PO across a wide spectrum of relative PO values elicited similar internal and external athlete responses, regardless of the direction travelled on a flowing river (i.e. upstream or downstream). The relationship between PO and S IMP during on-water sprint kayaking appears similar to those observed in rowing, where every percent change in boat speed measured relative to water (S IMP) requires a 2.87 and 2.94-fold percent change in paddling PO in female and male sprint kayak athletes, respectively. Continued evaluation of the PO-speed relationship for individual athletes may provide further insight into modelling performance and training targets for sprint kayak athletes.