54 subjects participated in a visual scanning study in which each subject was provided with only a single target set (of 1, 2, 4, 6, 8, or 10 letters in length). Eight sessions of 30 trials each were completed for each subject. Although there were slight differences in the rate at which performance improved over trials, this was not systematically related to size of target set. Moreover, even in the last session, there were large differences in performance in the different target sets. These findings suggest that Neisser's evidence for parallel preattentive processing in such tasks may have been confounded by his use of nested target sets and a within-subject design.