Nowadays rehabilitation robotics are common in rehabilitation. Traditionally these robots focus on the recovery of function (impairment) after for instance a stroke. When comparing conventional training with training using rehabilitation robotics literature shows no clear added value of the robotic devices. Although functional recovery of balance and gait following stroke may be quite substantial, little is known about the underlying mechanisms (‘determinants’) of such recovery in patients with different degrees of stroke severity. Such knowledge is mandatory to better focus rehabilitation strategies in patients with different degrees of stroke severity. Den Otter (2006) and Buurke (2008) showed that functional improvements after stroke may be more related to other mechanisms than to the restoration of muscle coordination patterns of the paretic leg. Furthermore, de Haart (2004) stated in her study that an important part of standing balance recovery from stroke takes place independent of improved dynamic contribution of the paretic leg. This talk will address the question whether or not rehabilitation robots should focus on improvement of function (impairments). Wouldn’t it be better to use the existing knowledge and technology to compensate for lost function instead?