Objective: We conducted a multisite, randomized, double-blinded, controlled trial to examine the effectiveness of a digital health intervention targeting the intrinsic regulation of goal-directed alertness in patients with chronic hemispatial neglect. Methods: Forty-nine participants with hemispatial neglect, who demonstrated significant spatially biased attention after acquired brain injury, were randomly assigned to the experimental attention remediation treatment or the active control group. The participants engaged with the remotely administered interventions for 12 weeks. The primary outcome was spatial bias on the Posner cueing task (response time difference: left minus right target trials). Secondary outcomes included functional abilities (measured via the Catherine Bergego scale and Barthel index), spatial cognition, executive function, quality of life, and sleep. Assessments were conducted before and immediately after participation in the experimental intervention or control condition, and again after a 3-month no-contact period. Results: Compared with the active control group, the intervention group exhibited a significant improvement in the primary outcome, a reduction in spatially biased attention on the Posner cueing task (p = 0.010, Cohen's d = 0.96), in addition to significant improvements in functional abilities as measured on the Catherine Bergego and Barthel indices (p = 0.027, Cohen's d = 0.24). Interpretation: Our results demonstrate that our attention training program was effective in improving the debilitating attention deficits common to hemispatial neglect. This benefit generalized to improvements in real-world functional abilities. This safe, highly scalable, and self-administered treatment for hemispatial neglect might serve as a useful addition to the existing standard of care. ANN NEUROL 2020;88:747–758.