The existing intracerebral hemorrhage (ICH) scores were based on the clinical and anatomical parameters of all primary ICH. We aimed to study whether the original ICH Score can predict cerebral amyloid angiopathy (CAA)-related ICH mortality and functional outcome and whether modified score can improve the predictions. The patients with ICH were consecutively recruited from 21 tertiary and secondary hospitals across Mainland China from January 2012 to December 2014. CAA-related ICH was defined as Boston Criteria. Logistic regression was performed in the derivation cohort of patients with CAA-related ICH to identify predictors of 3-month mortality and good outcome [modified Rankin score (mRS) of 0-2 at 3 months]. The areas under the receiver operating characteristic curves (AUCs) were used to assess model discrimination. A total of 360 CAA-related ICH patients were included. According to AUCs, the original ICH Score was less reliable predictor for mortality (AUCs=0.69) and good outcome (AUCs=0.67) in CAA- related patients. The range of CAA-related ICH score values is 0 to 7. The scale consist of four clinical items and the score points were assigned based on the Glasgow Coma Scale score on admission, age, presence of intraventricular hemorrhage, and presence of midline shift. CAA-related ICH score showed good discrimination in the derivation cohort (AUCs: 0.87 for mortality; 0.80 for good clinical outcome) and validation cohort (AUCs: 0.89 for mortality; 0.81 for good clinical outcome). The original ICH Score may be less reliable in predicting mortality and good clinical outcome at 3 months for CAA-related ICH patients. The modified scores improve its ability to predict clinical outcome at 3 months for CAA-related ICH.