When viewing the inside of a mask or mold of a human face, the face is frequently perceived as being a normal (convex) face, instead of the veridical, hollow (concave) face (the hollow face illusion). Thus, familiarity with the shape of faces dominates perception, even when in conflict with stereo depth cues. It has been suggested that visuomotor tasks are not affected by illusions that fool perception (Aglioti et al, 1995). In a previous experiment, we showed that reaching is affected by the hollow face illusion. However, the stimuli used lacked haptic feedback, which may produce fundamentally different (pantomimed) reaches which are driven by perception, not the usual visuomotor processes (Goodale et al, 1994). In work described here, we investigated whether the hollow face illusion holds for visuomotor tasks when haptic feedback is present. Computer images of normal and hollow faces were presented in stereo, such that stereo and familiarity depth cues were consistent or in conflict. In the visuomotor task, participants reached to either the nose or cheek. At the end of the reach, subjects received haptic feedback at the tip of the reaching finger. The maximum distance reached was used as an estimate of target position. In the perceptual task, subjects gave a numerical estimate of the distance to either the nose or cheek in arbitrary units, chosen by each subject. The perceptual and visuomotor distance estimates were similar. Both were dominated by object familiarity, shown by the nose estimates being closer to the subject than cheek estimates. However, hollow faces were estimated to be flatter than normal faces. This suggests that the visual system combines stereo and familiarity cues resulting in a 'flattening' of the hollow face. These results are consistent with the previous results from stimuli that did not include haptic feedback.