ABSTRACT A major problem facing both physicians and men with metastatic prostate cancer is predicting whether a specific therapy will be effective. Currently, when a man develops metastatic castration resistant prostate cancer, the initial therapy is frequently an inhibitor of androgen receptor activity such as enzalutamide or abiraterone acetate. Following progression on hormonal therapies, taxane-based chemotherapy is frequently employed. While these therapies improve overall survival and delay progression, there is great heterogeneity between patients in the chances of clinical benefit, as defined by response rates and durations of responses. While most men develop resistance to these second generation hormonal therapies within 1-2 years, some men derive many years of benefit, while others do not respond or respond only transiently. Thus, an unmet need is the ability to identify those men most likely to have durable benefits from these therapies, while sparing those men unlikely to benefit the costs and toxicities associated with these agents. Predictive biomarkers provide such an opportunity to optimize care delivery in this setting and reduce the heterogeneity of this disease. In addition, such biomarkers will permit the design of novel approaches and clinical studies designed to improve outcomes in those men in greatest need.