In a task requiring an optimal hand pointing (with regards to both time and accuracy) at a peripheral target, there is first a saccade of the eye within 250 ms, followed 100 ms later by the hand movement. However the latency of the hand movement is poorly correlated with that of the eye movement. When the peripheral target is cut off at the onset of the saccade, there is no correlation between the error of the gaze position and the error of the hand pointing. This suggests an early parallel processing of the two motor outputs. The duration of hand movement does not change significantly when subjects either see or not see their hand (closed or open loop). In the open loop situation, the undershoot of the hand pointing increases with target eccentricity, whatever the subjects are allowed or not to do a saccade toward the target. It suggests that the encoding of eye position by itself is a poor index for an accurately guided movement of the hand.