Rats were trained to perform a sustained attention task that required the subject to insert its head into a cylindrical "observation tunnel" and wait for the presentation of one of three spatially separated visual stimuli located on the upper portion of the tunnel circumference. Detection of a briefly (0.125 s) presented "correct" stimulus, followed by the rats' forward nose poke, resulted in access to a reinforcement dipper lifted through the orifice in the floor of the tunnel. Nose pokes to the two incorrect stimuli resulted in a 5-s time-out period. The task maximized attention and minimized movement requirements. Performance was characterized in terms of accuracy (i.e. errors of omission, and errors of commission), time on task, and latency to respond to the stimuli (i.e., reaction time). Haloperidol (0.04, 0.08, and 0.16 mg/kg) increased errors of omission and reaction time. However, lack of significant correlations between these two measures suggested that attentional accuracy may be independent of motor slowing produced by this neuroleptic. Scopolamine (0.2 mg/kg) alone increased both errors of omission and commission, but did not affect reaction time to correct stimuli. The sustained attention task as implemented here may be useful in the simultaneous study of classical neuroleptics desirable and undesirable CNS effects.