To determine the sensitivity, positive predictive value, and interobserver variability of CT in the detection of bullae associated with spontaneous pneumothorax in dogs. Retrospective case series. 19 dogs with spontaneous pneumothorax caused by rupture of bullae. Dogs that had CT for spontaneous pneumothorax caused by rupture of bullae confirmed at surgery (median sternotomy) or necropsy were included. Patient signalment, CT protocols, and bulla location, size, and number were obtained from the medical records. Computed tomographic images were reviewed by 3 board-certified radiologists who reported on the location, size, and number of bullae as well as the subjective severity of pneumothorax. Sensitivities of the 3 readers for bulla detection were 42.3%, 57.7%, and 57.7%, with positive predictive values of 52.4%, 14.2%, and 8.4%, respectively, with the latter 2 readers having a high rate of false-positive diagnoses. There was good interobserver agreement (κ = 0.640) for correct identification of bullae. Increasing size of the bulla was significantly associated with a correct CT diagnosis in 1 reader but not in the other 2 readers. Correct diagnosis was not associated with slice thickness, ventilation protocol, or degree of pneumothorax. Sensitivity and positive predictive value of CT for bulla detection were low. Results suggested that CT is potentially an ineffective preoperative diagnostic technique in dogs with spontaneous pneumothorax caused by bulla rupture because lesions can be missed or incorrectly diagnosed. Bulla size may affect visibility on CT.