Background: Studies have shown that certain Internet interventions can help alleviate depression. However, many such interventions contain personal support elements, making it difficult to ascertain whether the program or the support drives the effects. Studies are needed to investigate whether Internet interventions contribute to symptom reduction even when they are delivered without personal support, and even among severely depressed individuals who often receive other forms of treatment. Objective: This randomized controlled trial aimed to examine the effect of an Internet intervention that was deployed without personal support ("Deprexis") among adults with initially severe depression symptoms. Methods: Adults recruited from a range of sources who had exceeded the threshold for severe depression (PHQ-9 ≥ 15) in a pre-screening assessment and met inclusion criteria were randomized (N = 163) to the intervention (3 months program access; n = 78) or care-as-usual/waitlist control (n = 85). A diagnostic screening interview was administered by telephone at baseline to all participants. Online assessments were administered at baseline, 3 months (post-treatment), and 6 months (follow-up). The main outcome was the Patient Health Questionnaire (PHQ-9) between baseline and post-treatment. Results: Eighty-two percent of randomized participants were reached for the post-treatment assessment. Results for the intention-to-treat (ITT) sample showed significant intervention effects on depression reduction between baseline and post-treatment (linear mixed model [MM], F 1,155.6 = 9.00, p b .01, for the time by condition interaction), with a medium between-group effect size, Cohen's d = 0.57 (95% CI: 0.22-0.92). Group differences in depression severity at follow-up were marginally significant in the ITT sample, t (119) = 1.83, p = 0.07, and smaller than at post-treatment (PHQ-9, d = 0.33, 95% CI: −0.03-0.69). The number needed to treat (NNT) at post-treatment was 5, with 38% of participants in the intervention group achieving response (at least 50% PHQ-9 symptom change, plus post-treatment score b 10), compared to 17% in the control group, p b 0.01. Effects on secondary outcomes, including anxiety, health-related quality of life, and somatic symptoms, were not significant, with the exception of significant effects on anxiety reduction in PP analyses. Early ratings of program Internet Interventions 2 (2015) 48-59 ⁎ Corresponding author atInternet Interventions j o u r n a l h o m e p a g e : w w w . i n v e n t -j o u r n a l . c o m / helpfulness/alliance (after 3 weeks) predicted pre-post depression reduction, controlling for baseline severity and early symptom change. Conclusions: These results replicate and extend previous findings by showing that Deprexis can facilitate symptomatic improvement over 3 months and, perhaps to a lesser degree, up until 6 months among adults with initially severe depression.