Appropriate RT field in preoperative CRT for locally advanced thoracic esophageal carcinoma is still controversial. We investigated retrospectively a cohort of patients with clinically staged T3 to T4 with intention to treat (ITT) preoperative CRT followed by surgery in order to clarify the significance of ENI related to tumor location. We started a cohort study from 2011 for clinically staged T3 to T4 thoracic ESCC with preoperative CRT (50.4Gy/28F and concurrent 2 courses of FP) followed by planned surgery with a fixed 3-field lymph node dissection. When the 3-field ENI was used, lung constrains were V5<55%, V10<37%, and V20<25% to keep risk of Grade 3-5 radiation pneumonitis (RP) less than 4%. After 2015 we adopted the concepts of involved field (IF) or ENI related to tumor location without any necessary 3-field ENI. This change in fields setting has made it possible to analyze retrospectively whether ENI affects prognosis. The prognostic factors were derived from 161 patients, and 131 of them in fact were treated with surgery 1.5 months after CRT. In 30 patients without surgery, 14 experienced CR after CRT, and other 13 and 3 were inoperable because of tumor progression and toxicities, respectively. After ENI setting became to be optional, a few regulations were as follows: when tumor location was Ut, the upper end of ENI was No104 level, when tumor location was Mt or Lt, the upper end was No101 level. Five-year overall survival (OS) in 161 patients (average 65 years old) was 50.9%. There was no significance in 5-Y OS between each 2 groups in cT3 vs. cT3/4 (T3 undeniable T4 and T4) and in cN0 vs cN1-3 (T3: 59 vs. T3/4:102; p=0.11, cN0:15 vs cN1-3:146; p=0.09). In the single variate analyses, gender (male: 127 vs. female: 34; p=0.012), surgery (CRT-S: 131 vs. CRT alone: 30; p=0.009), and upper mediastinal ENI (UM-ENI)(UM-ENI+: 123 vs. ENI-: 38; p=0.009 were significant. These 3 prognostic factors were also significant in the multivariate analysis and each hazard ratio (Exp.β) and p-value were as follows; female: (3.29) p=0.002, surgery: (2.31) p=0.003, and UM- ENI: (1.44) p=0.012. On the other hand, abdominal ENI (A-ENI)(A- ENI+: 133 vs. ENI-: 28)was not significant (p=0.076) and neither in 79 patients whose tumor location was Lt (A-ENI+: 67 vs. ENI-:12, p=0.067). To find out serious toxicities which affected OS, causes of 70 deaths were analyzed excluding 3 unrelated deaths. Six (3.7%) died of the adverse events, and 3 out of them died of RP. As all of them were treated with UM-ENI, 3 out of 123 (2.4%) treated with UM-ENI died of RP. Clinical relevance of the UM-ENI in preoperative CRT was suggested in this study, because of the object limited in locally advanced stages and with a constant surgical method. Furthermore, clinical evaluation of the UM-ENI combined and RP risk should be done with a randomized controlled trial.