Background: Internet-delivered cognitive-behavior treatment for insomnia (iCBT-I) has the potential to fill the gap created by the discrepancy between insomnia cases and number of trained professionals. Although the effectiveness of this method was proven in multiple studies conducted in research settings, its feasibility in routine care is still unclear. Predictors, mediators, and moderators of treatment effect remain uncertain since previous studies often give contradictory results. The present study aims to investigate clinical effectiveness and cost-effectiveness of an internet-based CBT-I program Sleepsy in comparison with care as usual (CAU) among patients with CI recruited from clinical settings. Baseline data will be further analyzed to find predictors of treatment outcome. Methods/Design: The proposed study is a parallel-group randomized controlled trial comparing CAU plus iCBT-I with CAU in a clinical setting. One hundred ten participants will be referred from the medical doctors in Moscow. Both groups will have access to CAU, which corresponds to the treatment prescribed by the referring doctor. Patients of the first group will additionally get access to the iCBT-I program with the opportunity to contact a specialist (guidance on request) in a secured environment. The primary outcome is insomnia severity change from pre-to post-treatment. Secondary outcomes include change of subjective sleep characteristics, life quality, fatigue, daytime sleepiness, comorbid affective disorders, dysfunctional beliefs about sleep, sleep hygiene, healthcare consumption, productivity losses, and longer term outcomes at 3 months follow-up. Predictor analysis will include baseline scores of the aforementioned outcomes along with treatment expectancies and personality traits. Discussion: The proposed study is one of the first studies evaluating whether iCBT-I also works in routine care. We expect that recruitment of the participants let us determine the target group more precisely and exclude health problems interfering with treatment. Using CAU as control condition may result in a loss of power to detect a meaningful difference.