Abstract. The purpose of this work was to evaluate the effect of reconstruction algorithms, reconstruction parameters and definition of region of interest (ROI) on the accuracy of standard uptake value (SUV) quantification. Measurements were performed on a Siemens Biograph mCT PET/CT using a Jaszczak phantom which contains spheres of different diameter filled with two sphere-to-background activity ratios. All data were reconstructed using FBP, OSEM, OSEM+PSF, OSEM+TOF and OSEM+PSF+TOF reconstruction algorithms. The variation of reconstruction parameters such as iteration numbers and FWHM of Gaussian filter was also evaluated. Different ROI definitions were used to measure SUV in each sphere: SUVave, SUVmax, SUV50 and SUVpeak. SUV accuracy was estimated with recovery coefficients (RC) for all settings. An increase in RCs of 40.7 ± 1.2% from sphere diameters of 10 mm to 28 mm and an increase of 6.3 ± 1.2% in RC values was obtained when PSF was applied. It was observed that RCs increased when iteration number increased; RCs decreased when FWHM increased. In ROI definition, SUVmax yielded an important overestimation of activity concentration (near 20%) when PSF was applied. SUV50 yielded similar values for all algorithms used, except for FBP. SUVpeak yielded values similar to the real ones (SUV~1) for all reconstructions, particularly when tumor size was > 17 mm. In conclusion, it was observed that the use of a higher iteration number, lower FWHM and the use of PSF reconstruction algorithm improved accuracy of SUV quantification. SUV50 and SUVpeak provided a more accurate quantification than SUVmax and could be proposed to replace it.