Objective To explore the role of prostate-specific antigen (PSA) combined with the biopsy Gleason score (GS) in predicting the metastatic risk in patients with newly diagnosed prostate cancer. Methods The imaging and clinical data of 85 untreated prostate cancer patients, who were newly diagnosed by prostate biopsy and underwent ⁶⁸gallium-labelled prostate-specific membrane antigen (PSMA) ligand PSMA-11 (⁶⁸Ga-PSMA-11) positron emission tomography-computed tomography (PET-CT) in Changhai Hospital of Naval Medical University (Second Military Medical University) from Jan. to Dec. 2019, were analyzed retrospectively. There were 46 cases without metastasis (non-metastasis group) and 39 cases with metastasis (metastasis group). A logistic regression model for predicting the metastatic risk was established using the occurrence of ⁶⁸Ga-PSMA-11 PET-CT positive metastasis as a binary dependent variable and PSA×(GS-5) as a continuous independent variable, and the diagnostic efficacy was analyzed by receiver operating characteristic (ROC) curve; the model was tested with external validation data (n=20) to evaluate the accuracy of regression equation in predicting tumor metastasis. Results The median PSA (75.0[7.7-533.9]ng/mL vs 13.83[3.8-62.0]ng/mL, P<0.01) and GS (8[6-10]vs 8[6-10], P=0.042) in the metastasis group were significantly higher than those in the non-metastasis group. When PSA×(GS-5) was used as a continuous independent variable, the area under the ROC curve of predictive value was 0.857 (95% confidence interval[CI] 0.772-0.942, P<0.01). The best cut-off value of PSA×(GS-5) was 130.62, with a sensitivity of 71.8%, a specificity of 95.7%, and a Yoden index of 0.675. The regression equation logit (P)=0.019×PSA×(GS-5)-2.3 had an accuracy of 81.2% (69/85) for predicting the metastatic risk. There was no significant difference between the external validation data and the modeling data in the PSA (Z=-1.616, P=0.106) or GS (Z=-1.391, P=0.164). The external validation data met the test conditions, and the accuracy of the test regression equation was 85.0% (17/20). Conclusion PSA combined with GS has good performance in predicting the metastatic risk of patients with newly diagnosed prostate cancer.