The goal of our investigation was to study the effectiveness of the corrective reconstruction methods in cardiac SPECT using a realistic phantom and to qualitatively and quantitatively evaluate the reconstructed images using Bull's-eye plots. A 3D mathematical phantom which realistically models the anatomical structures of the cardiac-torso region of patients was used. The phantom allows simulation of both the attenuation distribution and the uptake of radiopharmaceuticals in different organs. Also, the phantom can be easily modified to simulate different genders and variations in patient anatomy. Two-dimensional projection data were generated from the phantom and included the effects of attenuation and detector response blurring. The reconstruction methods used in the study included the conventional filtered backprojection (FBP) with no attenuation compensation, and the first-order Chang algorithm, an iterative filtered backprojection algorithm (IFBP) and the ML-EM algorithm with non-uniform attenuation compensation. The transaxial reconstructed images were rearranged into short-axis slices from which Bull's-eye plots of the count density distribution in the myocardium were generated. The FBP reconstructed images clearly demonstrated the effects of attenuation on the myocardial distribution caused by anatomical structures and the inadequacy of the algorithm to provide accurate quantitative reconstructions. The images reconstructed using the Chang, IFBP, and ML-EM methods with attenuation compensation showed substantial improvement in terms of reduction in artifacts and improvement in quantitative accuracy. The results also demonstrate that the ML-EM method is the most robust in attenuation compensation for a wide range of anatomical variations when compared to the Chang and IFBP methods. The combination of the 3D mathematical phantoms and Bull's-eye plots provides an effective tool for evaluation of corrective reconstruction methods for quantitative cardiac SPECT.