The purpose of this study was to compare (18)F-FDG PET and PET/CT in a population of patients with colorectal cancer. PET and PET/CT images from 45 patients (17 women, 28 men; mean age +/- SD, 60.8 +/- 11.1 y) with known colorectal cancer referred for PET from June to November 2001 were retrospectively reviewed. Images were acquired with a PET/CT scanner, and (68)Ge attenuation correction was applied. PET images and fused (68)Ge attenuation-corrected PET and CT images were independently and separately interpreted by a moderately experienced reader unaware of the clinical information. Certainty of lesion characterization was scored on a 5-point scale (0 = definitely benign, 1 = probably benign, 2 = equivocal, 3 = probably malignant, 4 = definitely malignant). Lesion location was scored on a 3-point scale (0 = uncertain, 1 = probable, 2 = definite). The presence or absence of tumor was subsequently assessed using all available clinical, pathologic, and follow-up information. Analysis was provided for lesions detected by both PET and PET/CT. The frequency of equivocal and probable lesion characterization was reduced by 50% (50 to 25) with PET/CT, in comparison with PET. The frequency of definite lesion characterization was increased by 30% (84 to 109) with PET/CT. The number of definite locations was increased by 25% (92 to 115) with PET/CT. Overall correct staging increased from 78% to 89% with PET/CT on a patient-by-patient analysis. PET/CT imaging increases the accuracy and certainty of locating lesions in colorectal cancer. More definitely normal and definitely abnormal lesions (and fewer probable and equivocal lesions) were identified with PET/CT than with PET alone. Staging and restaging accuracy improved from 78% to 89%.