Although behavioral therapies are effective for posttraumatic stress disorder (PTSD), access for patients is limited. Attention-bias modification (ABM), a cognitive-training intervention designed to reduce attention bias for threat, can be broadly disseminated using technology. We remotely tested an ABM mobile app for PTSD. Participants (N = 689) were randomly assigned to personalized ABM, nonpersonalized ABM, or placebo training. ABM was a modified dot-probe paradigm delivered daily for 12 sessions. Personalized ABM included words selected using a recommender algorithm. Placebo included only neutral words. Primary outcomes (PTSD and anxiety) and secondary outcomes (depression and PTSD clusters) were collected at baseline, after training, and at 5-week-follow-up. Mechanisms assessed during treatment were attention bias and self-reported threat sensitivity. No group differences emerged on outcomes or attention bias. Nonpersonalized ABM showed greater declines in self-reported threat sensitivity than placebo (p = .044). This study constitutes the largest mobile-based trial of ABM to date. Findings do not support the effectiveness of mobile ABM for PTSD.