The aim of this study was to investigate insomnia symptom severity as a putative mediator of treatment response in therapist-guided internet-based cognitive behavioral therapy (ICBT) for chronic stress, using data from a randomized controlled trial. Participants (N = 100) were randomized to 12 weeks of ICBT or to a waitlist control condition (WLC). Insomnia severity was assessed weekly with the Insomnia Severity Index (ISI), as were the stress-related outcomes the Perceived Stress Scale (PSS) and the Shirom-Melamed Burnout Questionnaire (SMBQ). Latent growth models indicated that ICBT (vs. WLC) predicted a decrease in insomnia severity (alpha-path), and that growth in insomnia severity was predictive of growth in perceived stress and exhaustion (beta-paths). Most importantly, there were also significant indirect effects (alpha beta products) such that the beneficial effects of ICBT on perceived stress and exhaustion were mediated by a reduction in insomnia symptom severity (PSS: alpha beta =-0.44, 95% CI [-0.92,-0.14]; SMBQ: alpha beta =-0.08, 95% CI [-0.15, 0.04]). Explorative analysis of moderated mediation showed that more severe insomnia symptoms at baseline were associated with larger mediated effects. We conclude that reducing insomnia severity could be of importance for achieving successful treatment outcomes in ICBT for chronic stress.