Technological innovation has coincided with rapid improvements in performance in kayak sprint. With breakthroughs in materials, sensors, and wireless telemetry, instrumented equipment has unlocked new methods of quantifying and analysing performance. While some research has identified key biomechanical factors associated with performance, there is limited information available from on-water paddling and from new instrumented paddle devices. Thus, this thesis will improve the understanding of kayak performance through the examination of a novel instrumented paddle device. The device, developed by High Performance Sport New Zealand (HPSNZ), is capable of measuring blade force and velocity during normal on-water training conditions and may be adapted for ergometer use. A multi-level paddle protocol was created in partnership with coaches and support staff in order to replicate training and race intensities with maximal specificity. In Chapter 3, a study was performed to examine the validity and comparative reliability of the smart paddle (SP) relative to a popular kayak ergometer (DS). The SP and DS were practically identical in detecting stroke rate (SR) (limits of agreement = 0.02 ± 9.02%; R2 = 0.98; p < 0.01), but there were detectable differences in pull time (TPull) (limits of agreement = 10.1 ± 18.4%, R2 = 0.78, p < 0.01) and peak force (FPeak) (limits of agreement = 8.8 ± 30.1 N, R2 = 0.94, p < 0.01). Regardless, cyclical power variables were similar between SP and DS (SP IR and DS power; R2 = 0.98, SE = 0.045, p < 0.01) across all intensities. Chapter 4 uses the SP to compare kinetic and kinematic variables between on-water and kayak ergometer paddle environments. Large significant differences in TPull (d = 5.9 ± 0.39), air time (TAir) (d = 3.7 ± 0.27), mean force (FMean) (d = 1.06 ± 0.19), peak force (FPeak) (d = 1.92 ± 0.22), Impulse (d = 2.62 ± 0.23), and impulse rate (IR) (d = 2.10 ± 0.21) were found between environments. Kinetic differences expanded at higher intensities, which were visually apparent in statistical parametric mapping (SPM) analyses. Notably, IR was quite similar at maximal intensity (d = 0.28 ± 0.27). In Chapter 5, the previous results and other literature were used to examine correlations between performance variables and boat speed. The strongest correlations and predictive power for kayak velocity (VKayak) were with IR (R2 = 0.98, SE = 0.31, z = 1.86, p < 0.01) and cycle power (PCycle) (R2 = 0.95, SE = 0.035, z = 2.08, p < 0.01). Allometric scaling increased the predictive power of most kinetic relationships. Strong correlations were observed between DPS and FPeak (r = 0.69 ± 0.10), FMean (r = 0.68 ± 0.11), peak power (PPeak) (r = 0.72 ± 0.10), and impulse (r = 0.65 ± 0.18). Paddling efficiency (ep) was estimated between 0.65-0.75, on average, for all intensities. These data expand the body of knowledge surrounding kayak sprint biomechanics, suggesting that specific performance variables can predict performance and detect differences between athletes, paddling intensity, and environments. Instrumented paddle devices are a powerful tool with more potential to be explored.