Background Respiratory motion is one of the leading causes for artifacts in cardiac SPECT. We present a compensation method based on the knowledge of the respiratory state of the patient during list-mode acquisition. Methods: This method is based on the acquisition of list-mode SPECT data in synchronization with a respiratory signal acquired using pneumatic bellows. To test the correction methodology an anthropomorphic phantom of the human torso was manually moved to simulate respiratory motion during both emission and transmission acquisitions. The list-mode emission data were sorted into 1 of 8 possible respiratory states based on pneumatic bellows and rebinned so that one set of projections was obtained for each respiratory state. Then, at each projection angle, the 8 projections were summed after they were shifted so that the center-of-mass along the body axial direction was kept constant. Transverse slices were obtained using OS-EM with attenuation compensation. A clinical trial of this method has commenced. Results: With a 3-cm motion in the axial direction, the left ventricle is ovoid, with a significantly decreased apparent activity in the anterior and inferior walls. A 1.5-cm motion results in similar, but subtler, artifacts. Conclusion The method has been shown to significantly improve image quality in phantom acquisitions. We plan to develop a method for motion compensation applicable to list-mode transmission data and incorporate EKG gating in addition to respiratory motion compensation.