In this study evidence is given on the mechanisms subjects use to guarantee the stability of their social representations. An initial interview was held in which subjects were asked about the causes of AIDS, affected groups and modes of transmission. Their responses were found to divide into two social representations of AIDS: a conservative-blaming representation and a liberal representation. Subjects were then given technical information about AIDS. Two weeks later, the same subjects were given a free recall test and a recognition test. The results of these tests confirmed subject's tendency to select information which is congruent with their pre-existing representations, and to distort the recall of contradictory information in order to make it compatible with pre-existing representations.