Background: Lower level of albumin was related to worse prognosis of stroke and clinical trials showed that albumin therapy reduced mortality. However, stroke is heterogeneous and differences in the baseline concentration of albumin among subtypes of stroke were not assessed. The aim was to assess albumin level in patients with ischemic stroke classified by mechanism. Methods: Prospective controlled clinical study, including 200 patients with ischemic stroke and 50 controls. Patients were classified following Trial of ORG 10172 in Acute Stroke Treatment criteria. Plasma levels of albumin, fibrinogen, D-dimer, and C-reactive protein were assessed during 48 hr after admission. The National Institutes of Health Stroke Scale (NIHSS) on admission, in-hospital mortality, and Rankin score on discharge were recorded. Dependence was defined as mRS>2. Results: Patients with cardioembolic stroke showed significantly higher D-dimer and lower albumin. Mortality was related to higher NIHSS, higher D-dimer, lower albumin, and cardioembolic aetiology. Dependence was strongly related to lower albumin and higher NIHSS. Logistic regression: The cardioembolic aetiology (OR 0.101, 95 CI 0.0101.007, p .051) and the higher NIHSS score (OR 0.871, 95 CI 0.7581.002, p .053) were related to mortality; NIHSS (OR 1.560, 95 CI 1.3231.838, p < .0001) and older age (OR 1.052, 95 CI 1.0121.093, p .010) were independently related to dependence. Discussion: Patients with cardioembolic stroke showed lower albumin and higher risk of mortality than non-cardioembolic ones. Lower mean level of albumin was related to mortality and dependence in all patients. Reduced albumin may be a marker of chronic systemic inflammation, which may be the mechanism for cardiopathy and bad outcome of stroke. In addition, direct effects on ischemic tissue were suggested in experimental models.