The prognosis of prostate cancer (PCa) is related to tumor metastasis, among which 80% were bone metastasis. In this study, we investigated the correlation between diverse clinical factors and bone metastasis in PCa patients and identified potential biomarkers of bone metastasis in PCa patients. The clinical data of 150 PCa patients were reviewed consecutively from January 2015 to March 2020 in this study. The relationships between clinical characteristics, serum biomarkers and bone metastasis in PCa patients were analyzed, respectively. Multivariate logistic regression analysis was applied to identify potential markers of bone metastasis in prostate cancer. Receiver operating characteristic (ROC) curve was used to explore the diagnostic values of potential biomarkers. Compared with the PCa patients without bone metastasis, the serum levels of CA-125, T-PSA, F-PSA, CYFRA21-1 and ProGRP were significantly elevated in PCa patients with bone metastasis. Multivariate logistic regression analysis showed that T-PSA (OR 1.014, P = 0.021), F-PSA (OR 1.124, P = 0.016) and Pro-gastrin-releasing peptide (ProGRP) (OR 1.057, P = 0.026) were significantly associated with the bone metastasis of PCa patients. ROC curves indicated that T-PSA, F-PSA and ProGRP could effectively discriminate bone metastasis from non-bone metastasis PCa patients, and the AUCs (area under the curves) were 0.885, 0.919 and 0.752, respectively. According to the Youden index, the cut-off values of T-PSA, F-PSA and ProGRP were defined as 56.50 ng/ml, 6.96 ng/ml and 31.60 pg/ml, respectively. T-PSA, F-PSA and ProGRP produced a sensitivity of 78.30%, 81.70% and 61.70%, a specificity of 93.30%, 88.90% and 82.20%, respectively. The AUC for the combination of T-PSA, F-PSA with ProGRP was 0.941 with 90.00% sensitivity, much better than that of any single biomarker or two biomarkers combinate. Serum ProGRP might be a potential tumor marker of bone metastasis in prostate cancer, which may contribute to the early diagnosis of bone metastasis when used alone or in combination with other commonly used biomarkers. Copyright © 2020 Elsevier B.V. All rights reserved.