This assistance apparatus for upper limbs was developed for patients who can control their fingers but they cannot lift up their arms themselves, for example myopathy and hemiplegic patients. The purposes of the research in this paper are as follows: 1) to design the simple shaped arm and make the easy control system taking into account the practical use of patients, 2) to confirm the decrease of muscle activity while using the apparatus as an ADL assistance device for myopathy patients, 3) to grasp the tendency of the cerebral activity while using the apparatus as a Neuro-Rehabilitation assistance device for hemiplegic patients. The mechanism of assistance is utilizes the differential gears to lose the weight and volume of the mechanical arm. That enabled us to configure three motors to drive two DOFs (Degrees of freedom) for the shoulder and one DOF for the elbow around the root of the mechanical arm. The arm can lift up to 4 kgf (39.2 N) at the tip of the extended arm, and each maximum angle velocity is 1.57 rad/sec. This arm has two support trays, for wrist and upper arm. Each tray is equipped with a pressure sensor at the contact point to the user's arm, and by using the measured result of these four sensors, the control computer can learn the status of the user's arms. Furthermore, to realize other ADL (activities of daily living) motions (for instance, eating, writing, putting on making up, wiping his/her face, and so on) themselves, we proposed to control the device using the targeted posture map for the mechanical arm. At first, various desired or necessary postures of the assisted arms for the equipped person are determined, and each posture is defined as a control target. Next, each target (which is relatively close) is mutually connected, and the map can be accomplished. However, most of user can use the device without watching the map, because each target in the map is connected with the input signal of the same direction, which the user wants to move his/her hand. To be able to choose the appropriate input for each patient, various input interfaces, for example, joy-stick, push buttons, sensor glove using bending sensors, and so on, are equipped. The muscle activity while using the device was measured, and compared the %MVC data between using the device or not. As a result, the activity decreased up to 60%, and the effectiveness of this device could be confirmed. Finally, to expand the usage of this apparatus to encompass Neuro-Rehabilitation as well, the cerebral activity while using the device for rehabilitation with a near-infrared spectroscopy (NIRS) was measured. Then the data from using the device or not, and input motion from a third person were compared. By using this device, the cerebral activity decreased especially when the target motion was complex. However, when the subject input the motion themselves, the cerebral activity increased more than the data is input by a third person, especially, according to the complexity of the target motion. Therefore, for use in Neuro-Rehabilit...