We investigated the role of [18F]FDG positron emission tomography/computed tomography (PET/CT) in evaluating ground-glass nodules (GGNs) by visual analysis and tissue fraction correction. A total of 40 pathologically confirmed ≥1 cm GGNs were evaluated visually and semiquantitatively. [18F]FDG uptake of GGN distinct from background lung activity was considered positive in visual analysis. In semiquantitative analysis, we performed tissue fraction correction for the maximum standardized uptake value (SUVmax) of GGN. Of the 40 GGNs, 25 (63%) were adenocarcinomas, 9 (23%) were minimally invasive adenocarcinomas (MIAs), and 6 (15%) were adenocarcinomas in situ (AIS). On visual analysis, adenocarcinoma showed the highest positivity rate among the three pathological groups (88%, 44%, and 17%, respectively). Both SUVmax and tissue-fraction–corrected SUVmax (SUVmaxTF) were in the order of adenocarcinoma > MIA > AIS (p = 0.033 and 0.018, respectively). SUVmaxTF was significantly higher than SUVmax before correction (2.4 [1.9–3.0] vs. 1.3 [0.8–1.8], p < 0.001). When using a cutoff value of 2.5, the positivity rate of GGNs was significantly higher in SUVmaxTF than in SUVmax (50% vs. 5%, p < 0.001). The diagnostic sensitivity of [18F]FDG PET/CT in predicting the malignancy of lung GGN was improved by tissue fraction correction and visual analysis.