PURPOSE: To evaluate the sensitivity and specificity of conventional CT of abdomen in the detection of esophageal varices, and to correlate CT grade of esophageal varices with prognosis and risk for bleeding. MATERIALS & METHODS: Both CT and endoscopy were performed in 100 patients. Endoscopy revealed that while 54 patients had varices, 46 did not. CT criteria of variceal grading were follows : (1) wall thickening of more than 5mm or irregular wall contour(grade I) ; (2) intraluminal protruding tubular structures with contrast enhancement(grade II) ; (3) confluent varices in the wall of esophagus or multiplied paraesophageal collaterals(grade III). CT were reviewed by three radiologists without reference to clinical and endoscopic data. RESULTS: Sensitivity and specificity of CT in the detection of esophageal varices were 80%, retrospectively. CT and endoscopic grades agreed with each other in 68% of patients, and there was high correlation between CT and endoscopy.(Gamma statistics, p=0.828). No history or endoscopic evidence of variceal bleeding was present on gradeI, but there was a high incidence on grade II(35%) and on grade III(50%)(MH Chi-Square, Ridit scores=50.561,p=0.000). CONCLUSION: Abdominal CT is useful in the detection of esophageal varices, and can predict the risk factors of bleeding in patients with chronic liver diseases.