High efficiency cadmium–zinc–telluride (CZT) cameras provide an opportunity to lower the injected activities of radiopharmaceuticals for single photon emission tomography (SPECT) myocardial perfusion imaging (MPI). The limits for reducing activities of thallium have not been determined, particularly in obese patients. After an injection of 0.7 megabecquerel (MBq) of thallium/kg, we collected an average 1.5 million counts for the 10-min acquisition in a pilot cohort of ten patients. After extrapolation, we reduced the administered activity to 0.5 MBq/kg to obtain the expected 1 million counts. We studied the image quality in 124 patients (86 men, 43 obese with body mass index over 30 kg/m²) referred for MPI. The quality of images was assessed by a number of recorded counts and visually by a four-grade scale (one-poor quality, four-excellent quality). In non-obese and obese patients, the average number of recorded counts was 1.1 vs. 1.07 million counts for the 10-min stress acquisition, 1.04 vs. 1.06 million counts for the 13-min rest acquisition, and the average quality score was 3.97 vs. 3.90, respectively (p = NS).The mean administered activity was 39.2 ± 7 MBq for non-obese and 48.7 ± 6 for obese patients (p < 0.0001), and the calculated effective dose was 4.0 ± 0.7 and 4.9 ± 0.6 mSv respectively (p < 0.0001). The ultra-low-dose thallium stress-redistribution protocol, including post-stress prone imaging, provides good quality of images with a low radiation burden, even in obese patients.