Dual time-point imaging has been proposed as a means of improving the accuracy of 2-[18F]fluoro-2-deoxy-D-glucose positron emission tomography (18F-FDG PET) for the diagnosis of malignant pulmonary nodules. The purpose of this study was to evaluate a dual time-point protocol that has a narrow time window between its initial and its delayed imaging sessions. All patients examined during a 16-month time period, either for the diagnosis of a radiographically indeterminate thoracic lesion or for the staging of non-small-cell carcinoma, were included in the study provided that they completed the dual-point protocol and had either biopsy evidence of malignancy, biopsy evidence of a benign condition involving the thoracic lesion of concern, or clinical and radiographic follow-up consistent with the absence of malignancy. The entire study population was further divided into a central subpopulation, whose index lesions were adjacent to or within the hilum or mediastinum, and a peripheral subpopulation, whose index lesions were non-central. The maximum standardized uptake value (SUV) was measured for each lesion, and various body surface areas (BSAs) and glucose corrections on the SUV were compared using discriminant analysis. BSA corrected SUVs for the initial (iSUV) and the delayed (dSUV) imaging sessions, along with their absolute difference (deltaSUV) and fractional difference (fSUV) were also compared using discriminant analysis and receiver operating characteristic (ROC) analysis. The study population consisted of 132 patients, of whom 81 had malignancy and 51 were classified as having a benign condition. Thirty-three index lesions were central and 99 were peripheral; 109 had visible uptake and 23 had such low uptake that they were not visible above background. The mean time (+/-SD) between initial and delayed imaging for the visible lesions was 31.1+/-9.4 min. With respect to the entire study population, the BSA replacement for body weight gave the best performance among the various SUV corrections examined. In addition, the BSA corrected delayed SUV (dSUV) gave a performance superior to either initial SUV (iSUV), absolute difference in SUV (deltaSUV) and fractional difference in SUV (fSUV) alone. Performance gains achieved by BSA correction and by dSUV appeared to derive primarily from the central subpopulation, thereby indicating that central lesions tend to behave differently to peripheral ones. For the central subpopulation, ROC analysis also demonstrated improved detection of malignancy from dual-point imaging. The best performance was achieved when the BSA corrected dSUV was at least 2.4, or when the fSUV showed at least a 5% increase from initial to delayed imaging. With the optimal combined dSUV/fSUV strategy, the area under the ROC curve was 0.99, as opposed to 0.96 for dSUV alone, or 0.93 for iSUV alone. The ability of 18F-FDG PET to discriminate between benign and malignant conditions of the central thorax can be improved by correcting the SUV for BSA and by increasing the 'incubation time' between 18F-FDG injection and imaging, or by performing narrow time-window dual-point imaging.