This paper presents a new tool for assessment and therapy in post-stroke upper-limb rehabilitation and a new wireless sensor technology to enhance rehabilitation robotics based on the ZigBee network of wearable Inertial Measurement Units (IMU) and Surface Electromyography (sEMG) sensor nodes. These sensor nodes will allow the measurement of kinematic and electrical muscle activity of patients in continuous therapy motion over all body segments as a Body Sensor Network (BSN). The IMU Sensor design was based on a direction-cosine-matrix DCM. The system validation was achieved with an optical motion tracking system in which cameras and IMU sensors recorded upper limb positions simultaneously during a standard gesture of reaching and grasping. The comparison between elbow flexion-extension angle in reaching and grasping movements obtained from both techniques shows equivalence. The analysis of IMU data signals for several movements demonstrates high repeatability intra and inter-subjects.