In order to construct virtual environments, most systems have used 3 D displays utilizing retinal disparity. However, the depth perception of virtual objects with such displays is known to differ from that of real objects, which degrades task performance. In this study, we developed a hand-held device employing tactile phantom sensation providing depth cues in addition to a visual 3 D display, in order to handle virtual objects. The distance in the depth direction between a virtual object and the device was mapped to the position of the phantom sensation. Operator task performance with this device moving and deforming wire-framed virtual objects was measured under four display modes providing various depth cues. It was shown that the effectiveness of the device was related to the depth cues in the 3 D visual display methods and that the device was effective at reducing the time taken to deform virtual objects.