The purpose of this study was to investigate the contribution of proprioceptive and visual information about initial limb position in controlling the distance of rapid, single-joint reaching movements. Using a virtual reality environment, we systematically changed the relationship between actual and visually displayed hand position as subjects' positioned a cursor within a start circle. No visual feedback was given during the movement. Subjects reached two visual targets (115 and 125 degrees elbow angle) from four start locations (90, 95, 100, and 105 degrees elbow angle) under four mismatch conditions (0, 5, 10, or 15 degrees). A 2 x 4 x 4 ANOVA enabled us to ask whether the subjects controlled the movement distance in accord with the virtual, or the actual hand location. Our results indicate that the movement distance was mainly controlled according to the virtual start location. Whereas distance modification was most extensive for the closer target, analysis of acceleration profiles revealed that, regardless of target position, visual information about start location determined the initial peak in tangential hand acceleration. Peak acceleration scaled with peak velocity and movement distance, a phenomenon termed "pulse-height" control. In contrast, proprioceptive information about actual hand location determined the duration of acceleration, which also scaled with peak velocity and movement distance, a phenomenon termed "pulse-width" control. Because pulse-height and pulse-width mechanisms reflect movement planning and sensory-based corrective processes, respectively, our current findings indicate that vision is used primarily for planning movement distance, while proprioception is used primarily for online corrections during rapid, unseen movements toward visual targets.