: media-1vid110.1542/5804909711001PEDS-VA_2017-4182Video Abstract BACKGROUND: Knowledge about how children die in pediatric hospitals is limited, and this hinders improvement in hospital-based end-of-life care. METHODS: We conducted a retrospective chart review of all the patients who died in a children’s hospital between July 2011 and June 2014, collecting demographic and diagnostic information, hospital length of stay, location of death, and palliative care consultation. A qualitative review of provider notes and resuscitation records was used to create 5 mutually exclusive modes of death, which were then assigned to each patient. Analysis included the calculation of descriptive statistics and multinomial logistic regression modeling. RESULTS: We identified 579 patients who were deceased; 61% were <1 year of age. The ICU was the most common location of death (NICU 29.7%; PICU 27.8%; cardiac ICU 16.6%). Among the 5 modes of death, the most common was the withdrawal of life-sustaining technology (40.2%), followed by nonescalation (25.6%), failed resuscitation (22.8%), code then withdrawal (6.0%), and death by neurologic criteria (5.3%). After adjustment, patients who received a palliative care consultation were less likely to experience a code death (odds ratio 0.31; 95% confidence interval 0.13-0.75), although African American patients were more likely than white patients to experience a code death (odds ratio 2.46; 95% confidence interval 1.05-5.73), mostly because of code events occurring in the first 24 hours of hospitalization. CONCLUSIONS: Most deaths in a children's hospital occur in ICUs after the withdrawal of life-sustaining technology. Race and palliative care involvement may influence the manner of a child's death.