Patients with visual extinction were tested on three tasks involving stimulus identification and localization. In the first experiment, in which patients were to identify and localize stimuli, they demonstrated high levels of contralesional omissions. This primarily occurred under conditions of double simultaneous stimulation (DSS), consistent with the character of extinction. In contrast, when patients had to simply localize or count stimuli in Experiment 2, their contralesional omissions were very low. Similarly, when patients were to identify stimuli without localizing them (Experiment 3) they again showed very low contralesional error rates. These results support the view (Baylis, Driver, & Rafal, 1993) that visual extinction may be greatest when information about location cannot be bound to information about identity.