The purpose of this study was to compare the calibration of PET scanners and their cross calibration to peripheral devices in a multicenter study.Twenty-three dedicated PET scanners were investigated, applying standardized protocols. To ensure exact determination of the activity used, dose calibrators were checked using (68)Ge standards.Nine of 19 and 11 of 20 scanners displayed an error of <5% in 3-dimensional and 2-dimensional acquisition modes, respectively. Four and 5 scanners displayed an error of 10% in 3-dimensional and 2-dimensional modes, respectively. All other scanners yielded errors of 5% to <10%. Because of hardware and software problems, the measurements performed on 1 scanner could not be adequately analyzed.An investigation of calibration is mandatory. Especially for quantitative analyses in clinical multicenter trials, identification of potentially miscalibrated scanners is necessary.