UNLABELLED The purpose of this study was to evaluate the accuracy of image coregistration of PET and CT (PET/CT) images in patients with lung lesions and the influence of the breathing protocol during CT. METHODS Seventy-five patients with a solitary and well-circumscribed pulmonary lesion (non-small cell lung cancer; size, 10-30 mm) underwent PET/CT on a combined scanner. CT was acquired during shallow breathing in 37 patients and during normal expiration (i.e., the level reached when the patient exhaled without forcing expiration and then held the breath) in 38 patients. The volume of interest of each lesion was defined separately on PET and CT images, and the geometric center of gravity (COG) was assessed. The distance of COGs between the PET image and the CT image was measured. All lesions were classified according to 4 lung regions: apical, peripheral, central, and lung base. The mismatch between COG(PET) and COG(CT) was compared between regions and patient groups using a 2-way ANOVA with the Bonferroni-Dunn test for post hoc comparisons. RESULTS The range of COG distance between PET and CT was 1.7-5.4 mm in the apex, 0.5-14.7 mm in the periphery, 0.7-5.9 mm centrally, and 2.9-11.3 mm in the lung base. The match between PET and CT was significantly better in patients who had the CT scan obtained during normal expiration than in patients who performed shallow breathing during CT scanning (P = 0.024). No reciprocal effects were found (interaction P = 0.76). The mismatch of lesions depends significantly on lung region (P < 0.0001). Post hoc analysis showed a significant difference between the upper 2 regions and the lower 2 regions (all P < or = 0.002) but not between the apex and the central region (P = 0.95) and between the peripheral region and the lung base (P = 0.15). The lesion size had no influence on the COG mismatch. CONCLUSION The match of lung lesions in coregistered PET/CT images is better when acquiring the CT scan during normal expiration. The coregistration accuracy is better in the upper and central parts of the lung. The normal expiration protocol is suggested to be superior to shallow breathing during CT scanning.