Despite being ischemic stroke a leading cause of death and functional disability, there are no other accurate tools to predict outcome of patients beyond clinical variables such as age and stroke severity. In this scenario, defining protein changes associated with acute ischemic brain damage might help to identify new biomarker candidates for stroke prognosis. By means of mass spectrometry-based proteomics, we identified 51 proteins which levels were altered in the infarcted area of the human brain after stroke. Among 8 selected protein candidates, circulating levels of gelsolin, dihydropyrimidinase-related protein 2 and cystatin A were independent predictors of poor outcome. Logistic regression models including these innovative biomarkers significantly improved the predictive value with respect to the only use of clinical variables in both discrimination and reclassification analyses. Our results indicate that early blood determination of these three biomarkers might predict outcome of patients and might help in decision-making processes related to ischemic stroke management.