L’objectif de cette recherche est de tester la stabilité ou la variabilité des dérivés conscients des mécanismes de défense évalués à l’aide du questionnaire de Bond (DSQ 40) dans deux situations différemment anxiogènes afin de les intégrer dans le modèle transactionnel du stress de Lazarus et Folkman. Pour tester l’instabilité intersituationnelle des dérivés conscients des mécanismes de défense, une population de 161 étudiantes a complété le DSQ 40 et le questionnaire d’anxiété-état de Spielberger (STAI-Y), une première fois, en novembre, c’est-à-dire en l’absence d’évaluation de la performance universitaire, puis, une seconde fois, le premier jour de la session d’examen de janvier. Le score d’anxiété-état est significativement plus élevé à la seconde session qu’à la première. La majorité des 20 dérivés conscients des mécanismes de défense sont significativement différents aux deux sessions. L’inconstance intersituationnelle des dérivés conscients des mécanismes de défense observée dans cette recherche semble compatible avec l’hypothèse en faveur de leur variabilité. The aim of this research was to test the stability or variability of conscious by-products of defense mechanisms, evaluated with the Bond Defense Style Questionnaire (DSQ 40) so as to integrate them as transactional variables of stress in the model of Lazarus and Folkman. One hundred and sixty-one college girls (first-year students in a French university) engaged in psychology studies participated in this research. They participated in two collective sessions: a first session took place in November in a non-stressful situation, and a second session during the January examination week. Two questionnaires were used twice in this experiment: (a) the Anxiety State Questionnaire of Spielberger (STAI-Y) which evaluates the anxiety-state and anxiety-traits; (b) the DSQ 40, which estimates 20 conscious by-products of defense mechanisms and distinguishes three factors corresponding to the mature, neurotic and immature styles according to the factorial analyses of Guelfi and al. Two hypotheses were formulated: (a) a first hypothesis postulates that an examination situation increases the State Anxiety score; (b) a second hypothesis postulates a variability of defense mechanisms assessed by Bond's questionnaire (DSQ 40). It verified that the scores of the Bond defensive style questionnaire are significantly different in the two sessions. The mean state-anxiety score is higher in the second session. The great majority of students are described as more anxious during the week of examinations than during the previous month. The first hypothesis is then validated. The situation of evaluation of the university performance appears as a situation more stressful than the courses period. The three defense mechanism factors vary in a significant way from the first to the second session. In the second session the scores in mature and neurotic defense mechanisms increased significantly whereas the average scores in immature defense mechanisms decreased significantly. The majority of the 20 defense mechanisms are not stable between both situations. Indeed, 17 conscious by-products of defense mechanisms differ significantly in both sessions and only three defense mechanisms show a certain stability. Eleven defensive styles (55 %) decreased, six increased (30 %) and three remained stable (15 %). The instability of conscious by-products of defense mechanisms between the two sessions was confirmed by correlations analyses. The results show that the instability between situations is not limited to three categories of defenses (mature, neurotic and immature): 17 out of 20 mean scores in defense styles differed significantly. Conscious by-products of defense mechanisms are, indeed, influenced by situational factors.