Objectives: Respiratory gating reduces respiratory blur in cardiac SPECT. It can be implemented as: i) equal amplitude gating (AG); ii) phase or equal time gating (TG); or iii) amplitude based equal count gating (CG). The goal of this study is to evaluate the effectiveness of these respiratory gating methods for different respiratory patterns in patients undergoing myocardial perfusion SPECT imaging. Methods: The respiratory traces of 1282 SPECT patients obtained from Vicon Motion Systems tracking of reflective markers on the chest and abdomen were reviewed and categorized as 4 different respiratory patterns, in the order according to their prevalence: i) unclassified respiration (UR); ii) regular respiration (RR); iii) periodic respiration (PR) and iv) respiration with apnea (AR). For each pattern, the list-mode data of 5 selected patients which underwent Tc-99m MIBI stress SPECT were re-binned into 7 respiratory gates using AG, CG and TG respectively, after removing 5% spurious counts based on the histogram. A preliminary OS-EM reconstruction was performed for all respiratory gates with full compensation for resolution, attenuation and scatter. An ellipsoidal volume-of-interest (VOI) was drawn on the reconstructed images to extract the heart, and a rigid-body registration was performed to determine the heart motion between each gate and the 4th as reference gate. The resultant deformation parameters were incorporated in a final reconstruction to correct motion. The maximum extent of respiratory motion, FWHM of an image profile across the LV wall as an index of impact of respiratory blurring, and box-plot of the NSD values calculated from a uniform VOI drawn in the right lung, were measured for the 3 gating schemes for each respiratory pattern.Results: Compared to TG, the detected motion extent in AG and CG was 72±37% and 53±41% larger for the UR pattern; 25±23% and 16±13% larger for the RR pattern; 216±99% and 165±65% larger for the PR pattern; and 408±29% and 504±30% larger for the AR pattern. The average FWHM of AG, CG and TG was 15.26±1.96 mm, 14.76±2.10 mm and 16.24±2.74 mm for the UR pattern; 13.59±2.38 mm, 13.66±2.23 mm and 13.99±2.10 mm for the RR pattern; 15.99±3.94 mm, 15.19±3.89 mm and 16.64±3.38 mm for the PR pattern; and 13.77±2.57 mm, 14.43±2.95 mm and 16.26±3.03 mm for the AR pattern. Box-plot showed the NSD distribution through all gates in different gating schemes and there was no obvious difference among RR. TG had the smallest interquartile range in irregular respiration, followed by CG, while AG method showed largest interquartile range, especially in the AR and UR patterns.Conclusions: As expected, there is no substantial difference for 3 gating schemes in RR. AG and CG showed substantially larger motion estimation than TG in the AR and PR patterns. For the AR pattern, AG showed the best ventricular wall de-blurring yet the noise varied a lot in different gates. For the PR and UR patterns, CG showed best wall de-blurring with a relative uniform noise distribution. TG showed inferior motion estimation and de-blurring among all schemes. Based on the current data, CG is a robust respiratory gating implementation method given the respiratory pattern of the patients is unknown before data acquisition. Processing of more clinical patient data are warranted to allow statistical analysis between different gating methods.