Subjects were given descriptions of a person's behavior with instructions to form an impression of the person. The first behaviors in the series had implications for one trait, and the last behaviors had implications for a second trait that differed in favorableness from the first. After receiving the first set of behaviors, some subjects were told that an error had been made and that the behaviors should be disregarded. Other subjects were told instead to disregard the last behaviors presented. To-be-disregarded behaviors that occurred first in the series had little influence on judgements of either the specific trait to which they pertained or judgements of the target's likeableness, although subjects could recall these behaviors quite well. In contrast, to-be-disregarded behaviors at the end of the series did have an influence on specific trait judgements of the target, although they were recalled relatively poorly. These and other results were accounted for in terms of the general model of person memory and social information processing proposed by Wyer and Srull.