The aim of this study was to compare the power provided by a recent ergometer with the power developed by the rower determined using mechanical sensors set on the same apparatus. Six rowers and six non-rowers performed a power graded test and an all-out start on an instrumented ergometer (Concept2 system, model D, Morrisville, VT, USA). Power values displayed by the ergometer were recorded with a specific software. A strain gauge placed near the handle and a position sensor installed on the chain allowed the calculation of the power developed by the rower. Power values provided by the ergometer were strongly correlated to those determined with a direct measurement and calculation of power. However, power values given by the Concept2 system were lower (- 17.4 to - 72.4 W) than those calculated using mechanical sensors. This difference in power measurements was lower at a steady pace and for rowers. The Concept2 system underestimates the power produced by the rower by approximately 25 W. This difference in power seems to be independent of the level of power developed but increases with variations in intensity and pace. The deletion of the first strokes following changes in power production allows to limit this phenomenon. According to the use of the power parameter in the experimental design, it could be appropriate to correct values provided by the Concept2 ergometer.