We have investigated pointing movements towards remembered targets after an intervening self-generated body movement. We tested to what extent visual information about the environment or finger position is used in updating target position relative to the body after a step, and whether gaze plays a role in the accuracy of the pointing movement. Subjects were tested in three visual conditions: complete darkness (DARK), complete darkness with visual feedback of the finger (FINGER), and with vision of a well-defined environment and with feedback of the finger (FRAME). Pointing accuracy was rather poor in the FINGER and DARK conditions, which did not provide vision of the environment. Constant pointing errors were mainly in the direction of the step and ranged from about 10 to 20 cm. Differences between binocular fixation and target position were often related to the step size and direction. At the beginning of the trial, when the target was visible, fixation was on target. After target extinction, fixation moved away from the target relative to the subject. The variability in the pointing positions appeared to be related to the variable errors in fixation, and the covariance increases during the delay period after the step, reaching a highly significant value at the time of pointing. The significant covariance between fixation position and pointing is not the result of a mutual dependence on the step, since we corrected for any direct contributions of the step in both signals. We conclude that the covariance between fixation and pointing position reflects (1) a common command signal for gaze and arm movements and (2) an effect of fixation on pointing accuracy at the time of pointing.