It is known that visual shape perception is distorted in depth, but it is unclear how this influences grasping. We measured how subjects grasp real cylinders that are placed at eye height under normal lighting conditions. The cylinders were 10cm tall with an elliptical base. One principle axis was always 5cm, whereas the other was varied between 2cm and 8cm in steps of 1cm. The cylinders were placed directly in front of the subject, at a distance of either 15cm or 45cm. Their orientation was varied from 0 to 150 in steps of 30 . Grasping performance was compared with that in a previous study in which the cylinders were placed well below eye height so that subjects had no difficulty judging their shape. We found that subjects are less accurate at matching the orientation of their hand to that of the cylinders when the targets are at eye height. This often led to unstable grasps. Moreover the maximum grip aperture was about 2cm larger when the cylinders were placed at eye height. Nonetheless, the correlation between the hand orientation halfway through the movement and the final hand orientation was already about 70%. The correlation between the grip aperture during the movement and the final grip aperture was only 30% when half the distance was traversed, but it was still only 60% when reaching the object. These results indicate that the grasping movement uses incorrectly specified pick-up locations on the cylinder's surface. Uncertainty makes the subjects increase their grip aperture, but we did not find evidence for increased online control.