PURPOSE: To assess the clinical usefulness of the Bosniak classification system for characterizing cysticrenal masses by computed tomography (CT). MATERIALS AND METHODS: The CT scans of 28 patients with 32pathologically proven cystic renal masses (17benign, 15 malignant) were reviewed retrospectively by tworadiologists, who reached a consensus. Both unenhanced and enhanced CT scans were obtained using thin sectionswith 5-mm or 7-mm collination in 26 and four patients, respectively. Renal masses were categorized using theBosniak classification system; category I and II masses were considered benign, and category IV, malignant. Massesin category IIF or III were considered malignant when their wall thickness was more than 3 mm, their inner wallwas irregular or nodular, their septae were irregular, nodular or thick (more than 2-3 mm), calcification wasextensive, nodular or thick (more than 2-3 mm), and/or solid portions were enhanced more than 10-20 Hounsfieldunit. They were then correlated with their pathologic findings. RESULTS: For cystic renal masses, the finalpathologic results were as follows : all seven category I lesions were benign, as were seven of eight category IIlesions ; seven of 11 category III lesions were malignant, as were all seven category IV lesions. Distinguishingcategory II and III lesions by the Bosniak classification was difficult in two cases. One malignant tumor wasinterpreted as benign, and two benign tumors as malignant. With regard to prediction of the malignancy of 32cystic renal masses, as seen on CT, sensitivity, specificity, and accuracy were 93%, 88%, and 91%, respectively. CONCLUSION: Overall, the Bosniak classification system is useful for the evaluation and management of cystic renalmasses; however, with regard to the distinction between category II lesions requiring conservative management andsurgical category III lesions, some difficulties still remain.