PURPOSE: To assess the diagnostic accuracy and limitations of double contrast esophagography in patients with superficial esophageal cancer, as compared with endoscopic, gross and microscopic findings. MATERIALS AND METHODS: In 43 patients with pathologically proven superficial esophageal cancer, the detection rate and diagnostic accuracy of double contrast esophagography and endoscopy were compared. The depth of invasion revealed by esophagography, and grossly and microscopically in resected specimens, was compared. RESULTS: The detection rate and diagnostic accuracy were, respectively, 86.0% and 76.7% for esophagography, and 100% and 95.3% for endoscopy. In addition, very different detection rates (54.6% and 100%, respectively) were noted for epithelial and mucosal lesions. In flat-type cases (0-IIb), esophagography showed limited ability to detect lesions, but the accuracy of this modality in predicting the depth of tumor invasion was relatively high (94.6%). CONCLUSION: In cases of superficial esophageal cancer, double contrast esophagography showed a lower detection rate and lower diagnostic accuracy than endoscopy, and this was especially so for epithelial and mucosal lesions. The modality was able, however, to reliably predict the depth of tumor invasion.