The present study examined the accuracy of proprioceptive localization of the hand using two paradigms. In our proprioceptive estimation paradigm, participants judged the position of a target hand relative to visual references, or their body's midline. Placement of the target hand was active (participants pushed a robot manipulandum along a constrained path) or passive (the robot manipulandum positioned participants' target hand). In our proprioceptive-guided reaching paradigm, participants reached to the unseen location of a hand; both the left and right hands served as the target hand and the reaching hand. In both paradigms, subjects were relatively good at estimating the location of each hand (i.e. relative to a reference marker or using a reach), with directional errors falling within 2 cm of the actual target location, and little variation across the workspace. In our proprioceptive estimation paradigm, biases when the target hand was passively placed were no larger than those made when the target hand was actively placed. Participants perceived their left hand to be more to the left than it actually was, and their right hand to be more rightward than it actually was, but with a similar error magnitude across target hands. In our reaching paradigm, participants' estimates of left hand location were deviated more leftwards than their estimates of right hand location, but showed a small but similar pattern of location-dependent reach errors across the two hands. Precision of estimates did not differ between the two hands or vary with target location for either paradigm.