We have observed that filtered backprojection may cause artifactual decreased myocardial wall uptake in the reconstructed images if the hepatic-to-cardiac activity ratio (HCR) in 99mTc clinical myocardial SPECT studies is sufficiently high (> 1). To quantitatively relate hepatic uptake to this phenomenon, a commercial chest and heart phantom was modified with the addition of a customized liver insert, which was filled with various concentrations of 99mTc to simulate HCRs of 0:1, 1:1 and 2:1. The phantom was imaged with a high-sensitivity, three-detector camera, low-energy, high-resolution (LEHR) collimation and 180 degrees noncircular orbits. Quantitative circumferential profile analysis of the reoriented SPECT images demonstrated artifactual inferior/inferoseptal maximal activity decreases of 17.8% and 46.2% for the 1:1 and 2:1 HCRs, compared to the 0:1 HCR. Hepatic scatter probably partly mitigates the decrease. Smoothing the projection data before reconstruction worsened the artifacts' severity. Using Butterworth filters of order 5 and cutoff frequencies of 0.1, 0.2 and 0.215 Nyquist (clinical standard) resulted in artifactual inferior wall activity decreases of 5%, 8% and 16%, compared to using the same filter with a cutoff of 0.3 for an HCR of 2:1. These data indicate that if count statistics are good and liver uptake is high, higher frequency cutoffs in pre-reconstruction filters may improve specificity in 99mTc-labeled myocardial perfusion SPECT studies.