Subjects read a series of behaviors with instructions to form an impression of the person who performed them. In some conditions, subjects were told after reading the behaviors that an administrative error had been made and that certain ones should be disregarded. If the behaviors that subjects were told to disregard were descriptively unrelated to the other behaviors in the series, their influence on trait judgments was greatest when they were presented last. If the behaviors to be disregarded were descriptively inconsistent with the remaining behaviors, their influence was greatest when they were first in the series but subjects were not told to ignore them until after the remaining ones were presented. If the to-be-disregarded behaviors implied the same trait as the remaining ones, they had an influence on trait judgments in both of these conditions. All of these effects were consistently greater when the trait implied by the behaviors to be disregarded was favorable. In fact, when behaviors implied an unfavorable trait, instructions to disregard them often led the behaviors to have a contrast effect on trait judgments. Subjects appeared to base their judgments on implications of the cognitive representations they formed of the person at the time information was first presented. They then adjusted these subjective judgments at the time they reported them to compensate for the influence they perceived the to-be-disregarded information to have, making relatively greater adjustments when this information was unfavorable than when it was favorable. An additional experiment provided further evidence of adjustment processes, but it indicated that subjects also base their judgments on a partial review of the information they have received. A general model of person memory and judgment proposed by Wyer and Unverzagt (1985) provided a satisfactory account of the results of these studies.