Radiotherapy error in breast cancer mainly comprises setup error and respiratory breast motion. An accurate measurement of the setup error can provide a clinical reference for breast cancer planning target volume (PTV) expansion and improve treatment accuracy. Setup error rates were measured in 31 patients with breast cancer who underwent im-age-guided radiotherapy between June and October 2021. Cone beam computed tomography (CBCT) was performed for each patient first three times of treatment, and then once a week until the end of treatment. Bone registration was performed between the CBCT images and the CT images of the treatment plan. Record the registration results and obtain the setup errors. Calculated the external boundary of the PTV according to the setup error. A total of 131 CBCT scans were obtained for 31 patients. The standard deviation of systematic error and random error on the X (left and right), Y (the head and foot), and Z (the abdomen and back) axes were 1.40 and 0.67 mm, 1.89 and 0.56 mm, and 1.68 and 1.16 mm, respectively. The maximum absolute values of the setup errors were 4.9 mm, 6.4 mm, and 8.7 mm, respectively. The errors within 5.0 mm accounted for 100%, 98.47%, and 93.89%, respectively. The incidence of setup errors < 3 mm on the X, Y, and Z axes were 88.55%, 79.39%, and 75.57%, respectively. Based on the outward expansion of the clinical target volume, the PTV theoretical boundaries of breast cancer were 3.98 mm, 5.11 mm, and 5.02 mm, respectively. In foam-immobilized radiotherapy for breast cancer, the actual setup error cannot be accurately reflected considering the empirical PTV external boundary. It is clinically significant to measure and analyze the setup error using CBCT to calculate the PTV external boundary and guide clinical target area delineation.