Context COVID-19 has forced the implementation in many countries of measures to limit its spread such as quarantine, confinement or lockdown. This pandemic crisis is therefore a prolonged period of chronic stress and, for some, with traumatic characteristics. To our knowledge, there are no studies conducted on the effect of repeated measures of confinement. Our research studies the impact of three successive lockdowns on post-traumatic stress in France. We hypothesize that mental health deteriorates with each successive confinement. More specifically, the prevalence of post-traumatic stress should increase with the repetition of confinements. Method Three samples of adults were drawn during the first weeks of each of the three confinements. Our final three samples are 1316 for the first confinement (T1), 591 for the second confinement (T2), and 474 for the third (T3). The samples have an average age of MT1=32.8 (SDT1=11.7), MT2=37.0 (SDT2=13.4) and MT3=38.3 years (SDT3=13.1). Regarding the male-female distribution, it was also similar across samples. Participants were recruited through social networks. They completed a set of questionnaires via Qualtrics including the PCL-5 which assesses the intensity of post-traumatic stress disorder (total) and its dimensions (reexperiencing, avoidance, mood alteration and hyperreactivity). Results Results indicate an increase in the means of all of the PTSD levels between the beginning of the pandemic (T1) and the following measurement times, with a particularly large effect size ranging from .189 to .373. Post-hoc analyses indicate that participants reported a significant increase in post-traumatic symptoms between the first and second confinements and then all of these symptoms stabilized between T2 and T3 (no significant change between the second and third confinements). Logistic regression showed that the time of measurement significantly predicted the proportion of participants suffering from a diagnostic level of post-traumatic stress (Chi2(2) = 181, p < .001). This prevalence increased significantly from 16.2% in T1 to 43.3% in T2 (OR=3.9, p<.001) and then stabilized at 43.3% in T2 and at 38% in T3 (OR=0.82, p=.109). Conclusion Analyses show that post-traumatic stress levels were particularly high during the first two confinements and continued during the third one. Many psychosocial issues are raised during this period of health crisis. These results will be discussed with regard to the restrictions proposed by the government and the perspectives of public actions in terms of support during the confinement but also following such a health crisis.