Purpose: To compare magnetic resonance imaging (MRI) and computed tomography (CT) in detecting intracranial extension of the patients with nasopharyngeal carcinoma (NPC) and to analyze the impact on treatment outcome.Materials and Methods: From December 1997 to December 2004, 587 patients with histologically confirmed NPC (Iess than or equal to stage IVA according to American Joint Committee on Cancer, sixth edition) and received therapeutic radiation dose who underwent CT and MRI for evaluation of primary tumor extension. AII o f the images were reviewed and assessed by two physicians independently. Cases with variable interpretation or disagreement between the observers, these images were reevaluated side by side and any discrepancy was resolved by consensus. The kappa test for inter-observer concordance was 0.841 for CT and 0.931 for MRI.Results: The detection rate of intracranial invasion by MRI was 62.7% and that by CT was 15.5% (p＜0.001). The group with no intracranial invasion detected by MRI or CT (group 1) had significantly higher overall survival rate than intracranial invasion detected by MRI but not by CT (group 2, p＜0.001) or intracranial invasion detected by both MRI and CT (group 3, p＜0.001). For group 2 patients, chemotherapy plus radiotherapy had higher overall survival rate than radiotherapy alone (p＜0.001). Group 2 and 3 had the odd ratio of 3.2 and 4.0 comparing to group 1 by Cox regression. According to different treatment modalities, patients treated by radiotherapy alone had odd ratio of 2.6 comparing to chemotherapy plus radiotherapy.Conclusion: MRI owned a higher intracranial invasion detection rate than CT. Patients with intracranial involvement should receive both chemotherapy and radiotherapy in order to improve overall survival, even those with only subtle intracranial invasion.