We compared the predictive ability of the 2014 and 2005 Gleason grading systems in 568 patients initially diagnosed with metastatic prostate cancer (PCa). Outcomes included the duration of castration-resistant prostate cancer-free survival (CFS) and overall survival (OS). Univariate analyses and log-rank tests were used to identify prognosis indicators and assess univariable differences in CFS and OS in Gleason score (GS) groups. Cox proportional hazards and area under the curves of receiver operator characteristics methods were used to evaluate the predictive efficacy of the 2005 and 2014 ISUP grading systems. Univariate analyses showed that the 2005 and 2014 grading systems were prognosticators for CFS and OS; both systems could distinguish the clinical outcome of patients with GS 6, GS 7, and GS 8-10. Using the 2014 criteria, no statistical differences in patient survival were observed between GS 3 + 4 and GS 4 + 3 or GS 8 and GS 9-10. The predictive ability of the 2014 and 2005 grading systems was comparable for CFS and OS (P = 0.321). However, the 2014 grading system did not exhibit superior predictive efficacy in patients initially diagnosed with PCa and bone metastasis; trials using larger cohorts are required to confirm its predictive value. To the best of our knowledge, ours is the first study to compare the 2005 and 2014 grading systems in initially diagnosed PCa with bone metastasis. At present, we recommend that both systems should be used to predict the prognosis of patients with metastatic PCa.