In respiratory-gated cardiac SPECT with amplitude binning, the acquired data counts can vary greatly both between respiratory gates and among acquisition angles within each gate, which would lead to artifacts in reconstruction if not properly accounted for. In this work, we investigate a joint statistical reconstruction approach in which the different respiratory gates are described with respect to a common reference gate and the variation in the acquired data is compensated according to the acquisition time at each acquisition angle and respiratory gate. In the experiment, we demonstrated this approach with both simulated NCAT imaging (for quantitative evaluation) and a set of clinical acquisition. The results show that the proposed approach can further improve the reconstruction accuracy of the myocardium when compared to a post-reconstruction motion-compensation scheme developed previously.