Cognitive Behavioral Therapy (CBT) is effective in reducing insomnia complaints, but the effects of self-help CBT have been inconsistent. The aim of this study was to determine the effectiveness of self-help for insomnia delivered in either electronic or paper-and-pencil format compared to a waiting-list. Participants kept a diary and filled out questionnaires before they were randomized into electronic (n = 216), paper-and-pencil (n = 205), or waiting-list (n = 202) groups. The intervention consisted of 6 weeks of unsupported self-help CBT, and post-tests were 4, 18, and 48 weeks after intervention. At 4-week follow-up, electronic and paper-and-pencil conditions were superior (p < .01) compared to the waiting-list condition on most daily sleep measures (Δ d = 0.29-0.64), global insomnia symptoms (Δ d = 0.90-1.00), depression (Δ d = 0.36-0.41), and anxiety symptoms (Δ d = 0.33-0.40). The electronic and paper-and-pencil groups demonstrated equal effectiveness 4 weeks after treatment (Δ d = 0.00-0.22; p > .05). Effects were sustained at 48-week follow-up. This large-scale unsupported self-help study shows moderate to large effects on sleep measures that were still present after 48 weeks. Unsupported self-help CBT for insomnia therefore appears to be a promising first option in a stepped care approach. © 2011 Elsevier Ltd.