BackgroundFunctional electrical stimulation (FES) can be used for compensation of foot-drop for post-stroke individuals by pre-programmed fixed stimulation; however, this stimulation seems no more effective than mechanical ankle foot orthoses.ObjectiveWe evaluated the metrological quality of inertial sensors for movement reconstruction as compared with the gold-standard motion capturing system, to couple FES with inertial sensors to improve dorsiflexion on the paretic side, by using an adaptive stimulation taking into account individuals’ performance post-stroke.MethodsAdults with ischemic or hemorrhagic stroke presenting foot-drop and able to walk 10 m, were included from May 2016 to June 2017. Those with passive ankle dorsiflexion < 0° with the knee stretched were excluded. Synchronous gait was analyzed with the VICON© system as the gold standard and inertial measurement units (IMUs) worn by participants. The main outcome was the dorsiflexion angle at the heel strike and mid-swing phase obtained from IMUs and the VICON system. Secondary outcomes were: stride length, walking speed, maximal ankle dorsiflexion velocity and fatigue detection.ResultsWe included 26 participants [18 males; mean age 58 (range 45–84) years]. During heel strike, the dorsiflexion angle measurements demonstrated a root mean square error (RMSE) of 5.5°; a mean average error (MAE) of 3.9°; Bland-Altman bias of − 0.1° with limits of agreement − 10.9° to + 10.7° and good intra-class correlation coefficient (ICC) at 0.87 between the 2 techniques. During the mid-swing phase, the RMSE was 5.6; MAE 3.7°; Bland-Altman bias − 0.9° with limits of agreement − 11.7° to + 9.8° and ICC 0.88. Good agreement was demonstrated for secondary outcomes and fatigue detection.