Court practice has long regarded the stability of testimony, i.e., its unchangeability in successive interrogations of a witness, as the basic feature of reliability. However, some research into the psychology of testimony suggests that the stability of testimony becomes the actual criterion of reliability only when it has been projected to certain psychological features of a witness. This study examined 60 juvenile witness-victims of sexual abuse to attempt to determine the relation between witnesses' personalities and the quantity and character of changes in their testimony, and the relation of these factors to the evaluation of testimony reliability of the juveniles. When evaluating the reliability of testimony in which changes occurred, the following individual properties are significant: intellectual level, accentuated fear of social appraisal and heightened self-criticism. A strong relationship between the introduction of changes in testimony and individual properties and situational factors suggests the significance of the way in which witnesses showing certain psychic properties are interrogated. Tables, reference