Résumé : Purpose: Insomnia, being a mental disorder, is best conceived as a network of symptoms. With the important increase in insomnia prevalence during the COVID-19 pandemic, our aim was to investigate how the structure of insomnia symptoms in the general population has changed due to the pandemic. We also looked at the directional dependencies of nightmares and of covid-and lockdown-related stress/anxiety and depression in insomnia. Patients and Methods: 5986 persons replied to our online questionnaire for the first wave and 2843 persons to our second wave questionnaire. Both questionnaires included the Insomnia Severity Index (ISI). Regularized Gaussian Graphical Models (GGM) and Bayesian Directed Acyclic Graphs (DAG) were estimated. Results: The pre-and peri-lockdown networks were equally strongly connected (first wave: S = 0.13, p = 0.39; second wave: S = 0.03, p = 0.67), but differed for the first lockdown regarding only six edges (M = 0.13, p < 0.001) and for the second lockdown only five edges (M = 0.16, p < 0.001). These symptoms all worsened during the lockdowns in comparison to before the pandemic (p < 0.001). The diurnal items of the ISI had the highest predictability and centrality values in the GGMs. Lockdown-related stress/anxiety influenced indirectly nightmares through covid-related stress/anxiety, lockdown-related depressive affect and mental fatigue. These reported feelings of stress/anxiety and depression showed an indirect impact on insomnia symptoms through mental and physical fatigue. Conclusion: Though the lockdown slightly intensified insomnia symptoms, it did not alter their network structure. Despite their differences, both GGMs and DAGs agree that the diurnal symptoms of the ISI, play an essential role in the network structure. Both methods confirm the need for emphasizing the cognitive/affective component in the treatment of insomnia (ie cognitive behavioral therapy).