The distinction between reliability and validity is critical in examining utrocular identification. Four experiments demonstrated that two cues that lead to reliable discrimination do not lead to valid identification. Experiment 1 showed that, in the condition in which a stimulation of the right eye produced a visual direction toward the right and a stimulation of the left eye toward the left, there was a preponderance of correct responses. In the condition in which a stimulation of the right eye produced a visual direction toward the left and the left eye toward the right, there was a preponderance of incorrect responses. Experiment 2 showed that covariation of responses with visual direction decreased when feedback was provided because subjects sought other cues. Experiment 3, which included binocularly deficient subjects, showed that a feeling-in-the-eye is associated with the eye stimulated by a greater change in luminance rather than the eye stimulated by the target stimulus. When the luminance change was greater in the target eye, the feeling led to reliably correct responses, but when the luminance change was greater in the nontarget eye, it led to reliably incorrect responses. Experiment 4 indicated that the proportion of correct responses covaries with the degree of change in the luminance of the nontarget eye. The responses varied from reliably incorrect identifications, through unreliable identifications, to reliably correct identifications. These findings are consistent with the idea that stimulation of either eye is "projected" to the cyclopean eye.