Abstract： Objective To evaluate the variance and the concordance between the tumor length measured by CT scans and that measured by surgical specimens in esophageal carcinoma. Methods Fiftytwo surgical specimens of the esophageal carcinoma were made into pathological giant section.The shrinkage ratio of tumor was calculated by comparing the length of the specimen fixed by formalin for 24 h and that measured during the operation.One hundred and thirty-seven patients with esophageal carcinoma underwent spiral CT scan before the surgery,and the length of the gross tumor volume was obtained.After the tumor length of the fixed specimen had been measured,the real tumor length in situ was calculated using the shrinkage ratio.Then the variance and the concordance between the tumor length in CT scans and that in situ were compared.Results The mean shrinkage ratio was 90%±10%.The mean tumor length in CT scans was longer than that in situ(5.8 am±2.4 cm vs 4.1 cm±1.8 cm,P=9.68,P=0.000).The concordance of the length measured by the two methods was 40.9%(56/137). Conclusions A certain variance existed between the tumor length in CT images and that computed from surgical specimen in esophageal carcinoma.The results of esophagography and endoscopy should also be referred to delineate the gross tumor volume of esophageal carcinoma.