Vision has long been considered as a single feedback system providing information about the static and the dynamic features of motor behaviors and of the environment where they take place. However, recent models for oculo-manual movements have included multi-channeling of visual cues (Goodale & Milner, 1992; Jeannerod, 1981; 1984; Paillard, 1980, 1982). According to Paillard's model, a kinetic system, mostly sensitive to dynamic cues, provides directional information of the movement in the rapid distance-covering phase, and a static system, highly sensitive to position cues, provides positional signal errors. The present experiment gives kinematic evidence for the significant contribution of vision during the initial phase of rapid pointing movements when this phase is under the control of the kinetic channel. Movements having directional requirements were more accurate (directional precision) when vision of the initial portion of the trajectory was available. Times-to-peak acceleration and velocity were all shorter and their respective amplitudes were generally higher when vision was available for the first third of the trajectory than when it was not. Further, vision of the entire trajectory did not yield better precision then when vision was available for the initial phase of the movements only. Overall, the data support the existence of two corrective visual feedback systems.