PURPOSE: The purpose of this study was to evaluate the accuracy of each method of counting ribs on chest CT and to propose a new method: the anterior approach with using the sternocostal joints. MATERIALS AND METHODS: CT scans of 38 rib lesions of 27 patients were analyzed (fracture: 25, metastasis: 11, benign bone disease: 2). Each lesion was independently counted by three radiologists with using three different methods for counting ribs: the sternoclavicular approach, the xiphisternal approach and the modified sternomanubrial approach. The rib lesions were divided into three parts for evaluation of each method according to the location of the lesion as follows: the upper part (between the first and fourth thoracic vertebra), the middle part (between the fifth and eighth) and the lower part (between the ninth and twelfth). RESULTS: The most accurate method was a modified sternomanubrial approach (99.1%). The accuracies of a xiphisternal approach and a sternoclavicular approach were 95.6% and 88.6%, respectively. A modified sternomanubrial approach showed the highest accuracies in all three parts (100%, 100% and 97.9%, respectively). CONCLUSION: We propose a new method for counting ribs, the modified sternomanubrial approach, which was more accurate than the known methods in any parts of the bony thorax, and it may be an easier and quicker method than the others in clinical practice.