Three methods of determining cardiac output were compared in six anesthetized dogs (23-28 kg) for the purpose of evaluating the ability of ultrafast computed tomography to measure cardiac output at three points of the circulation: pulmonary artery, left ventricle, and descending aorta. Computed tomography cardiac output is calculated by indicator dilution analysis of an iodinated contrast bolus time-density curve. Computed tomography cardiac output was compared to thermodilution and radioactive microsphere cardiac output methods. The results show very good agreement between thermodilution and computed tomography methods (r = .90) and good agreement between microsphere and computed tomography (r = .88). Comparison of computed tomography cardiac output measurements from the left ventricle and pulmonary artery was excellent (r = .99), as were measurements from the left ventricle and descending aorta (r = .97). This study also showed minimal interlevel scan-density response variability (3%) and minimal variability between experiments (7%). Therefore, it is concluded that cardiac output can be accurately measured at many points in the blood pool by ultrafast computed tomography.