The current study compared EMG, stroke force and 2D kinematics during on-ergometer and on-water kayaking. Male elite flatwater kayakers (n = 10) performed matched exercise protocols consisting of 3 min bouts at heart and stroke rates equivalent to 85% of VO_2peak (assessed by prior graded incremental test). EMG data were recorded from Anterior Deltoid (AD), Triceps Brachii (TB), Latissimus Dorsi (LD) and Vastus Lateralis (VL) via wireless telemetry. Video data recorded at 50 Hz with audio triggers pre- and post-exercise facilitated synchronisation of EMG and kinematic variables. Force data were recorded via strain gauge arrays on paddle and ergometer shafts. EMG data were root mean squared (20ms window), temporally and amplitude normalised, and averaged over 10 consecutive cycles. In addition, overall muscle activity was quantified via iEMG and discrete stroke force and kinematic variables computed. Significantly greater TB and LD mean iEMG activity were recorded on-water (239 ± 15 vs. 179 ± 10 µV. s, p < 0.01 and 158 ± 12 vs. 137 ± 14 µV.s, p < 0.05, respectively), while significantly greater AD activity was recorded on-ergometer (494 ± 66 vs. 340 ± 35 µV.s, p < 0.01). Time to vertical shaft position occurred significantly earlier on-ergometer (p < 0.05). Analysis of stroke force data and EMG revealed that increased AD activity was concurrent with increased external forces applied to the paddle shaft at discrete phases of the on-ergometer stroke cycle. These external forces were associated with the ergometer loading mechanism and were not observed on- water. The current results contradict a previous published hypothesis on shoulder muscle recruitment during on-water kayaking.