Background: At present there is no simple, accurate blood test that may be used to determine the severity of stroke or to predict mortality and morbidity in stroke patients presenting to emergency departments. Methods: Patients with stroke-like symptoms who presented to an emergency department of a university hospital in Hong Kong were recruited for the study. DNA extracted from patients' plasma was analyzed for the -globin gene with a fluorescent-based PCR test. The primary outcome measures were in-hospital and 6-month mortality and morbidity using the post-stroke modified Rankin Score. Results: Among the 88 consecutive patients recruited to the study, 70 (80%) had ischemic stroke, 11 (13%) had intracerebral hemorrhage, and 7 (8%) had transient ischemic attacks. Median plasma DNA concentrations taken within 3 h of symptom onset were higher in patients who died compared with those who survived at discharge (6205 vs 1334 kilogenome-equivalents/L; P ؍ 0.03). Among patients with NIH Stroke Scale scores >8, median plasma DNA concentrations were higher in patients who died compared with those who survived to 6 months (2273 vs 968 kilogenome-equivalents/L; P ؍ 0.002). Plasma DNA concentrations correlated with the volume of cerebral hematoma (r ؍ 0.66; P ؍ 0.03). Plasma DNA concentrations >1400 kilogenome-equivalents/L had a sensitivity of 100% and a specificity of 74.4% for predicting hospital mortality after stroke, and the area under the ROC curve was 0.89 (95% confidence interval, 0.80 -0.94). The adjusted odds ratio for plasma DNA concentrations predicting 6-month mortality was