Post-stroke patients tend to move their trunk forward to reach an object rather than extending their affected arm, which is detrimental to good recovery of the paretic arm. In order to avoid this maladaptive trunk compensation, an objective assessment is needed. In that context, learned non-use (LNU) of the paretic arm is the difference between what the patient is capable of doing with his paretic arm and what he actually does. LNU has been previously assessed with a costly movement analysis device (Zebris). However, the Kinect 2 (X-box) is an innovative movement sensor that costs a fraction of the price of Zebris (about 200 €). The aim of this study was to show that the results obtained by the Zebris and Kinect are comparable and to see if the Kinect is a valid alternative to Zebris.Four post-supratentorial stroke participants were asked to reach a cone placed in front of them at 80% of their arm length. The reaching movement was repeated 5 times with the paretic hand, then 5 times with the less-impaired hand. This sequence was first performed with the trunk free, then with the trunk restrained. LNU of the paretic arm was obtained from the difference of the amount of trunk compensation between the free trunk condition and the restrained trunk condition measured by the Zebris and Kinect systems simultaneously.The results of a pilot study showed that the Kinect determination of LNU were similar to those obtained by the Zebris.This LNU score measured by Kinect was similar to that measured by Zebris, so we consider Kinect to be accurate enough to measure maladaptive trunk compensation. This LNU score could also be implemented in Kinect video games stroke neurorehabilitation. In the near future, patients could benefit from using low cost Kinect system in their own homes for rehabilitation and to assess their progress.