The finding that in a patient with visual form agnosia (DF), the performance level varies in a visuomotor letter-posting task and a perceptual orientation matching task was considered as part of the evidence for the perception–action model (Milner and Goodale, 1995). In this study we examined an alternative interpretation of these findings. We specifically tested whether orientation matching and letter posting can be accomplished using different strategies. Sixteen neurologically intact participants were asked to either put cards of different sizes through a target slot of a certain orientation or to simply indicate the slot's orientation. Letter-posting was performed in three different conditions varying the amount of visual feedback available. Results show that some participants apply a strategy of obstacle-avoidance in the posting task. That is, they oriented the card such that the safety margin between the edges of the target and the card was increased. This tendency became stronger with increasing card size. In contrast, in the orientation matching task, the end-orientation of the card was unaffected by its size and closer to the slot's actual orientation. The findings suggest that posting and matching can be solved using different visuo-spatial information. The perception–action dissociation reported for these tasks in DF might therefore simply indicate a difficulty in computing visual orientation, an ability that is needed for successful orientation matching but not for posting.