Purpose:To analyze the dependence of a cu te esophagitis (AE) on the dose and volume of irradiated esophagus during concur rent chemoradiation therapy of NSCLC.Methods:39 NSCLC patients previously treated with concurrent ch emoradiation therapy were studied , a subset of 62 patients who were recruited e arlier in a randomized phase Ⅲ study designed to evaluate the effect of amifost ine. The radiotherapy regimen was 1.2 Gy/fraction, Bid to a total dose of 69.6 G y. In this study 19 and 20 patients were in the control arm and amifostine arm, respectively. The AE score, evaluated by the RTOG acute morbidity criteria, was collected for each treatment week and one month following radiotherapy. DVH of t he esophagus was computed. A multivariate random-effect logistic model was used to investigate the correlation between the incidence of AE and various dose-vo lume factors among these 39 patients. The variables analyzed included total tumo r dose (TD), the mean and maximum dose to esophagus, the volume of esophagus tre ated above certain dose, and the dose of the esophagus treated to certain volume .Results:Among the 39 patients studied, two patients of the amif ostine arm (11%) and six patients of the non-amifostine arm (30%) experienced g rade Ⅲ AE. Grade Ⅲ AE was first seen at week 4th after a cumulative tumor dose to 36 Gy. However after week 4, it did not increase significantly with dose. Th e cumulative TD and patient sensitivity were found to have the strongest and sta tistically significant influence on the risk of AE in all dose and volume factor s.Conclusions:The risk of severe (grade Ⅲ) AE was associated wit h the cumulative dose to tumor and intrinsic patient sensitivity. Due to the tim e course of the development of AE, the daily dose rate or the cumulative dose by the 4th week of the treatment may be more predicative for the risk of AE rather than the total dose from the entire treatment course. Current clinical practice in using the total dose to assess the normal-tissue toxicity may need to be ad justed to account for the onset time of the acute end-point.