Background: Internet-delivered interventions can effectively change health risk behaviors and their determinants, but adherence to these interventions once they are accessed is very low. Therefore, it is relevant and necessary to systematically manipulate website characteristics to test their effect on website use. This study focuses on user control as a website characteristic. Objective: To test whether and how user control (the freedom of choice to skip pages) can increase website use and knowledge gained from the website. Methods: Participants older than 18 years were drawn from the Dutch Internet population (in June 2011) and completed a hepatitis knowledge questionnaire. Subsequently, they were randomly assigned to three groups: (1) a tunneled version of the website with less user control; (2) a high user control version of the website where visitors had the freedom of choice to skip pages; and (3) a control group that was not exposed to the website. Participants completed (1) a questionnaire of validated measures regarding user perceptions immediately after exposure to the website (except for the control group), and (2) a hepatitis knowledge questionnaire after one week to test whether participants in the experimental groups only clicked through the website or actually processed and learned its content. Server registrations were used to assess website use. Analyses of covariance (ANCOVA) using all available data were conducted to determine whether user control increases website use. Structural equation models (SEM) using all available data were constructed to test how user control increases website use-a latent variable derived from number of pages visited and time on website. Results: Of the 1044 persons invited to participate, 668 took part (668/1044, 64.0%). One half of participants (332/668 49.7%) were female and the mean age was 49 years (SD 16). A total of 571 participants completed the one-week follow-up measure regarding hepatitis knowledge (571/668, 85.5%). The findings demonstrate that having less user control (ie, a tunneled version of the website) had a negative effect on users' perception of efficiency (F-1,F-452 = 97.69, P <.001), but a positive effect on number of pages visited (F(1,45)2 = 171.49, P <.001), time on the website (F-1,F-452 = 6.32, P = .01), and knowledge gained from the website (F-1,F-452 = 134.32, P <.001). The direct effect of having less user control appeared to surpass the effect mediated by efficiency, because website use was higher among participants exposed to the tunneled version of the website in comparison with those having the freedom of choice to skip pages. Conclusions: The key finding that visitors demonstrated increased website use in the tunneled version of the website indicates that visitors should be carefully guided through the intervention for future intervention websites.