PURPOSE: To determine computed tomography (CT) criteria for cardiomegaly. MATERIALS AND METHODS: We analyzed posteroanterior chest radiographs and CT scans of 200 patients (M:F=130:70, mean age 49 years old) that were performed on the same day. On plain radiographs, the cardiothoracic ratio (R) was calculated using a standard method. On CT, we measured the maximal cardiac width (D(c)) and the maximal thoracic width of a patient (D(t1)). A second thoracic width was measured at the same scan level of D(c). Thus, two cardiothoracic ratios were derived in one patient-D(c)/D(t1) (R(1)) and D(c)/D(t2) (R(2)). We analyzed the appropriateness of R(1) and R(2) in the diagnosis of cardiomegaly to establish criteria for the use of the cardiothoracic ratio (ROC curve). RESULTS: When cardiomegaly was defined as a value of R that was greater than 0.5, both R(1) and R(2) were useful indicators of cardiomegaly. For a cut-off value of 0.5 for the cardiothoracic ratio for cardiomegaly, the sensitivity of R(1) and R(2) was 84% and 68%, respectively, and the specificity of R(1) and R(2) was 72% and 86%, respectively. CONCLUSION: The cardiothoracic ratio on CT can be easily obtained by measurement of the maximal cardiac width divided by the maximal thoracic width at the same scan level. When the cardiothoracic ratio on CT is over 0.5, the presence of cardiomegaly can be suggested.