This study compared two SPECT cameras with cadmium-zinc-telluride (CZT) detectors to a conventional Anger camera with cardiofocal collimators for the assessment of left ventricular (LV) function in a phantom and patients. Methods: A gated dynamic cardiac phantom was used. Eighteen acquisitions were processed on each CZT camera and the conventional camera. The total number of counts within a myocardial volume of interest varied from 0.25 kcts to 1.5 Mcts. Ejection fraction was set to 33%, 45%, or 60%. Volume, LV ejection fraction (LVEF), regional wall thickening, and motion (17-segment model) were assessed. One hundred twenty patients with a low pretest likelihood of coronary artery disease and normal findings on stress perfusion SPECT were retrospectively analyzed to provide the reference limits for end-diastolic volume (EDV), end-systolic volume (ESV), ejection fraction, and regional function for each camera model. Results: In the phantom study, for each ejection fraction value, volume was higher for the CZT cameras than for the conventional camera, resulting in a decreased but more accurate LVEF (all P , 0.001). In clinical data, body-surface-indexed EDV and ESV (mL/m 2 ) were higher for one of the CZT cameras (Discovery NM 530c) than for the other (D-SPECT) or the conventional camera (respectively, 40.5 ± 9.2, 37 ± 7.9, and 35.8 ± 6.8 for EDV [P , 0.001] and 12.5 ± 5.3, 9.4 ± 4.2, and 8.3 ± 4.4 for ESV [P , 0.001]), resulting in a significantly decreased LVEF: 70.3% ± 9.1% vs. 75.2% ± 8.1% vs. 77.8% ± 9.3%, respectively (P , 0.001). Conclusion: The new CZT cameras yielded global LV function results different from those yielded by the conventional camera. LV volume was higher for the Discovery NM 530c than for the D-SPECT or the conventional camera, leading to decreased LVEF in healthy subjects. These differences should be considered in clinical practice and warrant the collection of a specific reference database.