Objectives To develop a score for assessing the risk of symptomatic intracerebral hemorrhage (sICH) in ischemic stroke patients treated with iv thrombolysis. Methods The score was derived from a cohort of acute ischemic stroke patients treated with iv thrombolysis (n=972) at the Helsinki University Central Hospital (1995-2008). The predictive value of parameters associated with development of sICH (ECASS-2 criteria) was evaluated, and the score was developed according to the magnitude of logistic regression coefficients. The accuracy of the model was evaluated with 1000 bootstrap replicates. To test the performance of the score, we calculated area under receiver operating characteristic curve (AUC-ROC). We calculated absolute risks and likelihood ratios of sICH per increasing score points. The score was validated externally in merged cohort from 3 Swiss cities (Lausanne, Basel, and Geneva) including 828 ischemic stroke patients treated with iv thrombolysis. Findings The SEDAN score (0 to 6 points) consists of blood S ugar (glucose) at baseline [8.1-12.0 mmol/L (145-216 mg/dL)=1; >12.0 mmol/L (>216 mg/dL)=2], E arly infarct signs (yes=1) and (hyper) D ense cerebral artery sign (yes=1) on admission CT scan, A ge (>75=1), and N IH Stroke Scale on admission (>9=1). Accuracy of the score based on 1000 bootstrap replicates was 93.1% (91.4%-94.5%). AUC-ROC was 0.73 (0.67-0.79; p<0.001). Absolute risk for sICH in patients with increasing score points was: 1.4%, 2.9%, 8.5%, 12.2%, 21.7%, and 33.3% for 0, 1, 2, 3, 4, and 5 score points, respectively. Likelihood ratios were 0.19 (0.06-0.59), 0.40 (0.21-0.76), 1.23 (0.85-1.79), 1.85 (1.26-2.72), 3.68 (2.10-6.45), and 5.66 (2.40-13.40), respectively. Absolute risks in external validation were similar (1.0%, 3.5%, 5.1%, 9.2%, 16.9%, and 27.8%, respectively), and AUC-ROC was 0.77 (0.71-0.83; p<0.001). Conclusions The SEDAN score reliably assesses the risk of sICH in iv thrombolysis treated ischemic stroke patients and can support clinical decision-making in the high-risk patients. External validation of the score in the three Swiss cohorts supports its generalizibility.