Several perceptual studies have shown that the ability to estimate the location of the arm degrades quickly during visual occlusion. To account for this effect, it has been suggested that proprioception drifts when not continuously calibrated by vision. In the present study, we re-evaluated this hypothesis by isolating the proprioceptive component of position sense (i.e., the subjects were forced to rely exclusively on proprioception to locate their hand, which was not the case in earlier studies). Three experiments were conducted. In experiment 1, subjects were required to estimate the location of their unseen right hand, at rest, using a visual spot controlled by the left hand through a joystick. Results showed that the mean accuracy was identical whether the localization task was performed immediately after the positioning of the hand or after a 10-s delay. In experiments 2 and 3, subjects were required to point, without vision of their limb, to visual targets. These two experiments relied on the demonstration that biases in the perception of the initial hand location induced systematic variations of the movement characteristics (initial direction, final accuracy, end-point variability). For these motor tasks, the subjects did not pay attention to the initial hand location, which removed the possible occurrence of confounding cognitive strategies. Results indicated that movement characteristics were, on average, not affected when a 15-s or 20-s delay was introduced between the positioning of the arm at the starting point and the presentation of the target. When considered together, our results suggest that proprioception does not quickly drift in the absence of visual information. The potential origin of the discrepancy between our results and earlier studies is discussed.