PURPOSE: To evaluate the image quality of retrospectively attenuation corrected Positron Emission Tomography (PET) scans used for gross tumor volume (GTV) delineation in lung cancer patients. MATERIALS AND METHODS: Data of 13 lymph node positive lung cancer patients were acquired on separate CT and PET scanners under free breathing conditions (for radiotherapy planning). First we determined a protocol for CT/PET registration. Second, we compared the image quality of attenuation-corrected PET images using positron transmission images and CT images, in terms of signal-to-noise ratio (SNR) and lesion-to-background ratio (contrast). RESULTS: The largest differences between manual and automatic CT/PET registration were found in the anterior-posterior direction with a mean of 1.8 mm (SD 1.0 mm). Differences in rotations were always smaller than 1.0 degrees . The attenuation-corrected images using CT showed a larger SNR (mean 30%, SD 17%) and larger contrast (mean 14.0%, SD 8.5%) compared to attenuation-corrected images using positron transmission. For lymph nodes, the mean contrast was 16% (SD 6.4%) larger. CONCLUSIONS: This study demonstrated that attenuation correction based on CT provides a better image quality for GTV delineation than when using positron transmission for attenuation correction. Retrospective attenuation correction of PET scans based on registered CT is a good alternative for a dedicated PET/CT scanner if a free-breathing CT is available, e.g., for radiotherapy planning, and allows the use of CT with diagnostic quality for attenuation correction.