In visuomotor adaptation, when subjects first make a reach there is a mismatch between actual and predicted sensory feedback. Subsequently they adapt to accurately reach the target, and update their predictions about sensory feedback which informs changes in hand localization. Additionally, our lab and others have shown that hand proprioception recalilbrates to visual feedback ("proprioceptive recalibration"). Here we quantify the contributions of updated predicted sensory consequences and recalibrated proprioception to hand localization. Ifthere is no discrepancy between predicted and actual visual feedback of the hand during reach training, then any changes in hand localization are due to proprioceptive recalibration. In experiment 1, our subjects trained to actively reach to a target with a 30° rotation. In experiment 2, they were exposed to visual and proprioceptive discrepancy only (cross-sensory error signal; Cressman & Henriques, 2010), but since the apparatus moved the participants' hands there was no prediction about visual consequences to update. Then we measured changes in hand localization, also with both active and passive placement of the adapted hand; i.e. with and without predictions. Hand localization changed substantially in all conditions. With active training, the change for passive placement accounted for two thirds of the change for active placement. As expected, passive training did not lead to a difference between active and passive localization, suggesting both reflect proprioceptive recalibration only. This shows that cross-sensory error signals affect motor performance and lead to proprioceptive recalibration, and thus seem to be an unappreciated aspect of motor learning.