The standardized uptake value (SUV) is a relative measure of tracer uptake in tissue used in <I8>^F-FDG PET. However, the quality of ordered subset expectation maximization (OS-EM) images is sensitive to the number of iterations, because a large number of iterations leads to images with checkerboard noise. The main advantage of data acquisition in the three-dimensional (3D) mode is the high sensitivity to better exploit the intrinsic spatial resolution and the lower injection dose given to patients. In the 3D mode, the scatter fraction is higher, and, for a given administered dose, the random fraction is higher than that in the two-dimensional mode, which implies that correction methods need to be more accurate. Moreover, in clinical oncology 18F-FDG PET studies, patients have a wide variety of body shapes and sizes, which may impact image statistics. Consequently, it is necessary to make constant the acquisition (true) counts. The purpose of this study was to optimize injection dose and acquisition time in consideration of body mass index (BMI) for 3D whole-body <18>^F-FDG PET. Methods: A dedicated PET scanner, SIEMENS ECAT EXACT HR^+, was used to scan images of clinical data. The injection dose for BMI of < 14-19, 19-22, 22-25, and 25<(kg/m^2) were, 92.5 MBq, 111.0 MBq, 129.5 MBq, and 148.0 MBq, respectively. The emission scan time per bed position for BMI of < 14-19, 19-22, 22-25, and >25 (kg/m^2) were, 120, 120, 180, and 240 sec, respectively. A total of 20 patient subjects were evaluated as to true counts per bin (T/bin) of sinogram data and measured activity concentrations for the region of interest in the liver section. Results: T/bin was stable using an optimized protocol that took into consideration the BMI for any type of body morphology. The overall coefficient of variation was 7.27% for radioactivity concentration. Additionally, Gaussian filtering (8 mm FWHM) after reconstruction by the OS-EM method provided stable SUV values even when the iteration number was increased 30 times over. Conclusion: Optimization of injection dose and acquisition time indicated that BMI was a clinically useful acquisition protocol for 3D whole-body <18>^F-FDG PET.