Objectives: Despite compelling evidence of lipoprotein(a) [Lp(a)] as a risk factor for ischemic stroke, its underlying mechanism remains unclear. Our aim is to investigate whether serum Lp(a) level is associated with the extent and location of cerebral steno-occlusive lesions, and with large artery atherosclerotic (LAA) stroke in Korean patients. Methods: We analyzed data prospectively collected over a 3-year period on consecutive patients with stroke or TIA. Based on an angiographic study, a total of 1012 patients were classified into four subtypes: non-cerebral stenosis (n = 654), intracranial stenosis (n = 198), extracranial carotid stenosis (n = 86), and combined intracranial and extracranial carotid stenosis (n = 74). Independent associations of Lp(a) levels with the extent and location of cerebral stenosis were evaluated, and Lp(a) levels of subtypes by the TOAST criteria were compared. Results: Lp(a) levels of LAA stroke were significantly higher than those of the other four stroke mechanisms. Patients with more advanced intracranial (p = 0.001) and extracranial carotid stenoses (p = 0.001) tended to have higher Lp(a) levels. In multiple regression analysis, the third Lp(a) quartile was the strongest risk factor for isolated intracranial (OR 3.36, 95% CI 1.77-6.37) or extracranial stenosis (OR 4.82, 95% CI 1.96-11.88), whereas the fourth Lp(a) quartile was the most powerful predictor for combined intracranial and extracranial carotid stenosis (OR 4.98, 95% CI 1.92-12.91). Conclusions: Our results indicate that greatly elevated Lp(a) levels are associated with LAA stroke and extensive burden of cervicocerebral steno-occlusive lesions, which might offer indirect evidence of proatherothrombogenic role of Lp(a) in ischemic stroke. (C) 2010 Elsevier Ireland Ltd. All rights reserved.