Ferroptosis induced by lipid peroxidation is closely related to cancer biology. Prostate cancer (PCa) is not only a malignant tumor but also a lipid metabolic disease. Previous studies have identified ferroptosis as an important pathophysiological pathway in PCa development and treatment, but its role in the prognosis of PCa is less well known. In this study, we constructed a nine-ferroptosis-related gene risk model that demonstrated strong prognostic and therapeutic predictive power. The higher risk score calculated by the model was significantly associated with a higher ferroptosis potential index, higher Ki67 expression, higher immune infiltration, higher probability of biochemical recurrence, worse clinicopathological characteristics, and worse response to chemotherapy and antiandrogen therapy in PCa. The mechanisms identified by the gene set enrichment analysis suggested that this signature can accurately distinguish high- and low-risk populations, which is possibly closely related to variations in steroid hormone secretion, regulation of endocrine processes, positive regulation of humoral immune response, and androgen response. Results of this study were confirmed in two independent PCa cohorts, namely, The Cancer Genome Atlas cohort and the MSK-IMPACT Clinical Sequencing Cohort, which contributed to the body of scientific evidence for the prediction of biochemical recurrence in patients with PCa. In addition, as the main components of this signature, the effects of the and genes on ferroptosis were evaluated and verified by and experiments, respectively. The above findings provided new insights and presented potential clinical applications of ferroptosis in PCa.