Objectives To assess diagnostic accuracy and added value of dual time point 18 F-FDG PET/CT after gastric distention using oral water in differentiating malignant from benign gastric lesions. Methods Patients (n = 30, 19 males, mean age 58.6 ± 16.4 years). All patients are known or suspected oncology patients. All patients underwent whole body 18 F-FDG PET/CT scan and 2 h delayed PET/CT abdominal images following oral water gastric distension. The best cut off values for early SUVmax (SUVmax1), delayed SUVmax (SUVmax2) and SUVmax2-SUVmax1 (ΔSUVmax) to differentiate benign from malignant lesions were set based on ROC analysis. Data analyzed included in addition; age, sex and 18 F-FDG uptake pattern in delayed images. Suspicious gastric lesions were correlated with biopsy in 18 patients (60 %) and with clinical and follow-up imaging ( 18 F-FDG PET/CT, CT or MRI) in 12 patients (40 %). Unpaired t -test was used to compare the mean deference in continuous variables between patients with gastric malignancy and those with benign gastric lesions. Fisher’s exact test was used to analyze categorical variables. Logistic regression analysis was performed to identify the most powerful factors to predict malignant lesions. Results Fifteen patients (50 %) had confirmed malignant gastric lesions. Patients with confirmed gastric malignancy were older (65 ± 13 vs 52 ± 17; p = 0.023) and had significantly higher mean ΔSUVmax (1.29 ± 1.76 vs −0.89 ± 1.59; p = 0.003). The mean SUVmax1 (6.99 ± 6.66 vs 5.31 ± 2.53; p = 0.367) and SUVmax2 (8.29 ± 7.41 vs 4.44 ± 3.34; p = 0.077) although both higher in patients with malignant lesions, they did not reach statistical significance. Sensitivity, specificity, PPV, NPV, and accuracy to detect malignant gastric lesions were highest for lesions with localized uptake pattern in delayed images post water oral contrast as well as for lesions with ΔSUVmax>0. Regression analysis revealed both variables as independent predictors for malignant lesions with odd ratios of 22.9 and 9.5 respectively and final model Chi-Square of 19.9 (p < 0.0001). The model correctly identified 12/15 (80 %) malignant lesions and 13/15 (86.7 %) benign lesions with 2 false positives confirmed as chronic active gastritis with helicobacter pylori and 3 false negatives including 1 signet ring gastric cancer and 1 low grade gastrointestinal stromal tumor (GIST), both with poor 18 F-FDG uptake. Conclusion Localized uptake pattern in delayed PET/CT images following gastric distention with oral water contrast as well as ΔSUVmax>0 are powerful independent variables to identify malignant gastric lesions with fairly high sensitivity and reasonable accuracy. Malignancies with inherently low 18 F-FDG avidity are the main cause of false negatives while active gastritis is th...