To reduce potential mis-registration from differences in the breathing pattern between two complementary PET and CT data sets, patients are generally allowed to breathe quietly during a dual-modality scan using a combined PET/CT tomograph. Frequently, however, local mis-registration between the CT and the PET is observed. We have evaluated the appearance, magnitude, and frequency of respiration-induced artefacts in CT images of dual-modality PET/CT studies of 62 patients. Combined PET/CT scans during normal respiration were acquired in 43 subjects using single- or dual-slice CT. Nineteen patients were scanned with a special breathing protocol (limited breath-hold technique) on a single-slice PET/CT tomograph. All subjects were injected with approximately 370 MBq of FDG, and PET/CT scanning commenced 1 h post injection. The CT images were reconstructed and, after appropriate scaling, used for on-line attenuation correction of the PET emission data. We found that respiration artefacts can occur in the majority of cases if no respiration protocol is used. When applying the limited breath-hold technique, the frequency of severe artefacts in the area of the diaphragm was reduced by half, and the spatial extent of respiration-induced artefacts was reduced by at least 40% compared with the acquisition protocols without any breathing instructions. In conclusion, special breathing protocols are effective and should be used for CT scans as part of combined imaging protocols using a dual-modality PET/CT tomograph. The results of this study can also be applied to multi-slice CT to potentially reduce further breathing artefacts in PET/CT imaging and to improve overall image quality.