Abstract: Objective To investigate 18F-FDG PET/CT manifestations of massive type active tuberculosis and lung cancer and the differential diagnosis of the two diseases based on 18F-FDG PET/CT findings. Methods We retrospectively collected the data from 74 patients with active tuberculosis and 64 patients with lung cancer, whose lesions presented as solid masses on CT. The demographic and clinical data of the patients, 18F- FDG PET characteristics including SUVmax, 18F- FDG uptake (higher than mediastinal blood pool or not), radioactive defect within the lesion, and the CT findings including the lesion size, signs of cavity, vacuoles, lobulation, smooth border, and mediastinal/lung window ratio (M/L ratio) of the lesions were analyzed. Univariate and multivariate analyses were used to compare the variables between the two groups, and a logistic regression model was established for differentiation of the two diseases. The diagnostic efficiency was evaluated by area under the receiver-operating characteristic (ROC) curve analysis. Results No significant differences were found in the quantitative index (SUVmax >2.5 or not) or in the qualitative index (uptake of lesion higher than mediastinal blood pool or not) in PET between massive type active tuberculosis and lung cancer (P>0.05). Univariate analysis revealed that SUVmax, 18F-FDG uptake of the lesion, age, lesion size, signs of cavity, or M/L ratio were not significantly different (P>0.05), but gender, signs of radioactive defect, vacuoles, smooth border and lobulation were significantly different (P<0.05) between the two diseases. Multivariate analysis showed that gender, signs of radioactive defect, smooth border and lobulation of the lesion were independent factors for discrimination of the two diseases (P<0.05). A risk prediction model for active tuberculosis was established based on logistic regression analysis: P=1/(1+e-x), X=-0.530+1.978×gender+3.343×radioactive defect +2.846×smooth border-2.116×lobulation. For diagnosis of active tuberculosis, the sensitivity, specificity, accuracy, positive predictive value and negative predictive value of this model were 78.4%, 92.2%, 84.8%, 92.1%, and 78.7%, respectively. Conclusion The combined analysis of gender, signs of radioactive defect, smooth border and lobulation of the lesions is useful for discriminating massive type active tuberculosis from lung cancer in the majority of the patients, whereas 18F-FDG uptake alone has only limited value for a differential diagnosis.