To explore the signature function of the tumor mutational burden (TMB) and potential biomarkers in prostate cancer (PCa), transcriptome profiles, somatic mutation data, and clinicopathologic feature information were downloaded from The Cancer Genome Atlas (TCGA) database. R software package was used to generate a waterfall plot to summarize the specific mutation information and calculate the TMB value of PCa. Least absolute shrinkage and selection operator (LASSO) Cox regression analysis was used to select the hub genes related to the TMB from the ImmPort network to build a risk score (RS) model to evaluate prognostic values and plot Kaplan-Meier (K-M) curves to predict PCa patients survival. The results showed that PCa patients with a high TMB exhibited higher infiltration of CD8+ T cells and CD4+ T cells and better overall survival (OS) than those with a low TMB. The anti-Mullerian hormone (AMH), baculoviral IAP repeat-containing 5 (BIRC5), and opoid receptor kappa 1 (OPRK1) genes were three hub genes and their copy number variation (CNV) was relatively likely to affect the infiltration of immune cells. Moreover, PCa patients with low AMH or BIRC5 expression had a longer survival time and lower cancer recurrence, while elevated AMH or BIRC5 expression favored PCa progression. In contrast, PCa patients with low OPRK1 expression had poorer OS in the early stage of PCa and a higher recurrent rate than those with high expression. Taken together, these results suggest that the TMB may be a promising prognostic biomarker for PCa and that AMH, OPRK1, and BIRC5 are hub genes affecting the TMB; AMH, OPRK1, and BIRC5 could serve as potential immunotherapeutic targets for PCa treatment.