Results of surgery in locally advanced esophageal carcinomas (T3 and/or N1) are disappointing. Concomitant chemoradiotherapy (RTCT) gave equivalent survival results in many phase II studies. Two randomized trials (French and German) compared surgery or additional RTCT after a first phase of RTCT. Both drew the same conclusions, that is surgery did not improve overall survival but increased postoperative mortality. However, local control was found better in the surgical arms, and in some subgroups, esophagectomy improved disease-free survival suggesting that some patients may benefit from surgery. After preoperative RTCT, absence of residual disease in the surgical specimen (pathological complete response) occurs in 15 to 30%; these patients underwent a radical surgery without any benefit but with high risk of morbidity and mortality. Nevertheless, it is still difficult to select this sub-population: CT-scan or endoscopy with biopsies have a low sensitivity and specificity. 18-FDG-PET-scan, performed after or during the RTCT, is able to increase sensitivity, but only preliminary results with small populations are available. No biological factor of chemoradiosensitivity (p53, NfkappaB, p21...) could predict who will respond or not. Another approach is to reserve surgery only to patients with a demonstrated local failure (salvage surgery) but the feasibility of this technique is still debated. Finally, local relapses are frequent after RTCT and optimisation of the current schedules is mandatory to improve oncologic results. Unfortunately, increasing the radiation dose did not improve local control and showed more toxicities. New drugs as taxanes, oxaliplatine, or targeted therapies are tested in on-going phase III trial.