CDATA[Whether original intracerebral hemorrhage (ICH) score can be used to predict clinical outcomes in patients with SMASH-U (structural vascular lesions, medication, cerebral amyloid angiopathy, systemic disease, hypertension, or undetermined) classification remains an open question. This study obtained data related to consecutive acute patients with ICH from 21 tertiary hospitals in China during January 2012 to December 2014. Using the SMASH-U method, patients were classified into 6 subtypes. Favorable functional outcome and mortality was obtained after ICH at the 3 months. We used logistic regression to evaluate the effectiveness of each risk model in predicting clinical outcome and under the receiver operating characteristic curves (ROC) to assess performance. A total of 3475 patients were included, the most common cause was hypertensive angiopathy (n=1279, 36.81%), followed by undetermined (n=1168, 33.61%), cerebral amyloid angiopathy (CAA) (n=507, 14.59%), structural vascular lesions (n=368, 10.59%), medication (n=96, 2.76%), and systemic disease (n=57, 1.64%). For good clinical outcome (mRS≤2), the ROC values of original ICH score were 0.781, 0.701, 0.718, 0.722, 0.788, and 0.771, while for the mortality in 3-month, the ROC values of original ICH score were 0.840, 0.734, 0.836, 0.722, 0.785, 0.820, and 0.734 according to SMASH-U pathogenic classification, respectively. The ability of original ICH score may be well differentiated among the 6 ICH pathogeneses. Thus, physicians should select different risk score according to different etiological ICH.