We describe experiments that test the hypothesis that changes in somatosensory function accompany motor learning. We estimated psychophysical functions relating actual and perceived limb position before, and after two kinds of motor learning: directional motor learning (learning to reach in the presence of novel forces applied by a robot), and non-directional learning (learning to reach quickly and accurately to visual targets, without forces). Following force-field learning, sensed limb position shifted reliably in the direction of the applied force. No sensory change was observed when the robot passively moved the hand through the same trajectories as subjects produced during active learning. Perceptual shifts are reflected in subsequent movements: following learning, movements deviate from their pre-learning paths by an amount similar in magnitude and in the same direction as the perceptual shift. After non­directional motor learning in the absence of forces, we observed improvements in somatosensory acuity following learning. Acuity improvement was seen only in the region of the workspace explored during learning, and not in other locations. No acuity changes were observed when subjects were passively moved through limb trajectories produced during active learning. Taken together, our findings support the idea that sensory changes occur in parallel with changes to motor commands during motor learning, and that the type of sensory change observed depends on the characteristics of the motor task during learning.