This study adopted a novel approach to relating nonhuman and human studies of anxiety and latent inhibition, by exploring the degree to which rats' Btemperaments^in relation to anxiety predicted the development of latent inhibition. It investigated whether anxiety levels in one situation (i.e., an elevated-plus maze) involving 38 intact, mature rats, could predict performance on a latent inhibition task (i.e., an animal model of attention), and, thus, reproduce findings from human studies. Rats were subjected to two tasks: a novel within-subject, appetitive stimulus pre-exposure procedure, and an elevated-plus maze task. In the stimulus preexposure task, non-reinforced exposure to a light led to facilitation of conditioning (perceptual learning) during the first 3 days, and to retardation of conditioning (latent inhibition) during the last 5 days. In the elevated-plus maze task, moderate levels of anxiety were observed. Regression analyses revealed that anxiety levels (plus maze) were a significant predictor of latent inhibition (stimulus pre-exposure). Measures of locomotor activity did not predict performance on the latent inhibition task. Rats with moderate levels of anxiety had better performance in the late inhibition task than animals with low levels of anxiety. These data and the methodology have implications for understanding nonhuman models of schizophrenia, and for the design of studies investigating these issues with nonhumans.