Objective : The aim of this study was to analyze the treatment results and prognostic factors in patients with massive cerebral infarction who underwent decompressive craniectomy. Methods : From January 2000 to December 2005, we performed decompressive craniectomy in 24 patients with massive cerebral infarction. We retrospectively reviewed the medical records, radiological findings, initial clinical assessment using the Glasgow Coma Scale, serial computerized tomography (CT) with measurement of midline and septum pellucidum shift, and cerebral infarction territories. Patients were evaluated based on the following factors : the pre- and post-operative midline shifting on CT scan, infarction area or its dominancy, consciousness level, pupillary light reflex and Glasgow Outcome Scale. Results : All 24 patients (11 men, 13 women; mean age, 63 years; right middle cerebral artery (MCA) territory, 17 patients; left MCA territory, 7 patients) were treated with large decompressive craniectomy and duroplasty. The average time interval between the onset of symptoms and surgical decompression was 2.5 days. The mean Glasgow Coma Scale was 12.4 on admission and 8.3 preoperatively. Of the 24 surgically treated patients, the good outcome group (Group 2 : GOS 4-5) comprised 9 cases and the poor outcome group (Group1 : GOS 1-3) comprised 15 cases. Conclusion : We consider decompressive craniectomy for large hemispheric infarction as a life-saving procedure. Good preoperative GCS, late clinical deterioration, small size of the infarction area, absence of anisocoria, and preoperative midline shift less than 11mm were considered to be positive predictors of good outcome. Careful patient selection based on the above-mentioned factors and early operation may improve the functional outcome of surgical management for large hemispheric infarction.