While characterization of lung lesions and staging of lung cancer with fluorine-18 fluorodeoxyglucose positron emission tomography (FDG-PET) is an established clinical procedure, a lower diagnostic accuracy of FDG-PET for diagnosis and staging of so-called bronchioloalveolar carcinoma (BAC) has been reported. Therefore, the accuracy of PET for diagnosing and staging of BAC was investigated. We studied 41 patients eventually found to have adenocarcinoma with a bronchioloalveolar growth pattern who were referred for characterization or staging of lung lesions with whole-body FDG-PET between January 1998 and March 2001: there were 11 males (27%) and 30 females (73%), with a mean age of 66.0+/-10.9 (range =44-84 years). Patients were imaged using ECAT EXACT or HR+ systems. All patients had non-attenuation-corrected scans, while transmission data for attenuation correction were also available for 12 patients (29%). PET correctly identified BAC in 41 of the 46 (89%) lesions and 39 of the 41 patients (95%). By pathology, 25 patients (61%) were found to have unifocal or nodular lesions; this pattern was correctly identified by PET in 20 patients (80%) and by CT in 18 (72%). PET correctly identified 7 (44%) of 16 patients (39%) who had multicentric or diffuse BAC, and CT identified 11 (69%). Of the 35 patients whose lymph node status was verified pathologically, PET was correct in 27 (77%) and CT in 24 (69%). PET missed 67% of the rare tumors that had a pure BAC pattern with no invasive component. It is concluded that the diagnostic performance of whole-body FDG-PET is similar in most patients with lesions with a BAC pattern and in other non-small cell lung cancer types. PET is less accurate in patients with rare BAC tumors that have no invasive component.