BACKGROUND: The purpose of this article is to report the overall survival (OS) outcome of patients with nasopharyngeal carcinoma (NPC) with local failure who received salvage treatment and to identify prognostic factors for OS. METHODS: Between January 1996 and December 2000, 2915 patients received primary radiotherapy (RT) with or without chemotherapy for nonmetastatic NPC. At a median follow-up of 3.1 years, 319 patients had developed local failure as the first failure, with or without synchronous regional/distant failure. OS was calculated from the start of primary RT. Univariate and multivariate analyses were performed to identify prognostic factors for OS in patients with isolated local failure. RESULTS: The T classification distribution of the local failure (rT classification) was as follows: 68 (21%) rT1 to T2a, 92 (29%) rT2b, 82 (26%) rT3, and 77 (24%) rT4. The rT classification was the same as the initial T classification in 82% of patients. Two hundred seventy-five patients (86%) had isolated local failure, and 232 (84%) of them did not have any distant metastasis or regional failure develop during follow-up. Salvage treatment was given to 200 patients (73%) with isolated local failure. One hundred fifty-nine patients (80%) received reirradiation (108 external beam RT [EBRT], 44 brachytherapy, and seven EBRT plus brachytherapy), 22 patients (11%) underwent nasopharyngectomy with or without postoperative RT, and 19 patients (9%) were treated with chemotherapy alone. Four patients died of RT complications, and one died of chemotherapy toxicity in the absence of active NPC. The 3-year actuarial OS for patients with isolated local failure was 74%. On multivariate analysis, advanced initial T classification (hazard ratio [HR], 1.44; p = .0006) and the use of salvage treatment (HR, 0.54; p = .0038) were independent prognostic factors. For the subgroups of patients who had the same recurrent and initial T classification, salvage treatment was associated with improved OS only in the subgroup with T1 to T2 local failure (n = 127; p = 0.0446), but not in the subgroups with T3 (n = 48) or T4 (n = 54) disease. CONCLUSIONS: Most patients with first local failure have localized disease. Salvage treatment is feasible in most of the patients with clinically isolated local failure. Patients who had early initial T classification have a more favorable prognosis. Subgroup analysis suggests that salvage treatment only prolongs survival in patients with T1 to T2 recurrent disease.