Background: Suicide is a major public health problem, and it remains unclear which processes link suicidal ideation and plans to the act of suicide. Growing evidence shows that the majority of suicidal patients diagnosed with major depression or bipolar disorder report repetitive suicide-related images and thoughts (suicidal intrusions). Various studies showed that vividness of negative as well as positive intrusive images may be reduced by dual task (e.g. eye movements) interventions taxing the working memory. We propose that a dual task intervention may also reduce frequency and intensity of suicidal imagery and may be crucial in preventing the transition from suicidal ideation and planning to actual suicidal behaviour. This study aims a) to evaluate the effectiveness of an Eye Movement Dual Task (EMDT) add-on intervention targeting suicidal imagery in depressed patients, b) to explore the role of potential moderators and mediators in explaining the effect of EMDT, and c) to evaluate the cost-effectiveness of EMDT. Methods: We will conduct a multi-center randomized clinical trial (RCT) evaluating the effects of EMDT in combination with usual care (n = 45) compared to usual care alone (n = 45). Participants will fill in multiple online batteries of self-report questionnaires as well as complete a semi-structured interview (Intrusion Interview), and online computer tasks. The primary outcome is the frequency and intrusiveness of suicidal imagery. Furthermore, the vividness, emotionality, and content of the suicidal intrusions are evaluated; secondary outcomes include: suicidal behaviour and suicidal ideation, severity of depression, psychological symptoms, rumination, and hopelessness. Finally, potential moderators and mediators are assessed. Discussion: If proven effective, EMDT can be added to regular treatment to reduce the frequency and vividness of suicidal imagery. Trial registration: The study has been registered on October 17th, 2018 at the Netherlands Trial Register, part of the Dutch Cochrane Centre (NTR7563).