Current models of spatial attention are based on results from paradigms in which attention is allocated in two-dimensional displays. However, the distribution of attention in three dimensional space has received little consideration. A series of experiments investigated whether the covert orienting of visual attention task (COVAT) can be transferred into stereoscopic space. Participants viewed a visual display that consisted of two overlapping boxes centred on a fixation point. A polarising screen and software were used to induce the illusion that one box was in front of the fixation point while the other was behind. A cue was presented by brightening one of the boxes, subsequent targets could appear at either the cued or uncued location. In two experiments, targets were more likely to appear inside the cued box. In both detection and discrimination experiments, no significant differences were observed between valid and invalid targets. These data suggest that attentional selection in the COVAT is not allocated within a three dimensional representation of space.