The purpose of this study was to prospectively evaluate positron emission tomography (PET) for delineating lung cancers preradiotherapy and to assess PET's ability to distinguish residual tumor from scarring following radiotherapy. Between April 1991 and October 1992, 20 patients underwent 18fluoro-2-deoxyglucose (18FDG) PET scanning of the chest prior to radiotherapy for lung cancer. Tumor volumes on chest x-ray (CXR) and computerized tomography (CT) scan were correlated with abnormalities on PET scans. Follow-up PET studies were compared to postradiotherapy chest x-ray and/or CT scans, and correlated with clinical outcome. Six of seven well-demarcated tumors showed increased uptake of 18FDG correlating with the CT/CXR tumor volume. Twelve poorly demarcated tumors demonstrated increased 18FDG uptake. In seven of 12, the CT/CXR abnormality correlated with changes on PET scan. In three of 12, CT/CXR abnormalities were larger than on PET, whereas in two of 12, abnormalities on PET extended outside the region of CT/CXR changes. The 13th patient in the poorly demarcated category had diffuse carcinoma in situ at the surgical margin that demonstrates increased 18FDG uptake, but was not visible by CT/CXR. Of 12 patients with follow-up studies, all had changes on CXR and/or CT that made it difficult to assess response. Four of 12 had a complete response by PET; all remain locally controlled. The remaining eight patients had either a partial response (n = 6) or no response (n = 2) by PET. Four of these eight patients remain alive and well 11-24 months after therapy. 18FDG PET may be useful for delineation of lung cancer volumes that are poorly defined by CXR and/or CT scan. The value of PET in differentiating tumor from fibrosis after radiotherapy for lung cancer remains to be established.