Hormone-naïve prostate cancer and its castration-resistant state (CRPC) are clinically and genetically heterogeneous diseases. From initiation of prostate carcinogenesis to its evolution towards therapeutic resistance, various combinations of genetic and epigenetic events occur.Schematically, progression to CRPC could be divided in two distinct pathways, either dependent or independent of the androgen receptor activity. Nevertheless, because the better knowledge of the genetic landscape of CRPC is under way, limited clinical applications are available at the moment, underlying the usefulness of prognostic and predictive biomarkers in daily practice. Despite the promising prognostic value of circulating tumor cells, no biomarker has been currently validated as a surrogate for overall survival in CRPC patients.Inversely, considerable interest has been generated with the recent finding of the splice variant AR-V7 that allows to predict resistance to abiraterone acetate and enzalutamide.However, other predictive biomarkers would be necessary to accurately guide personalized sequencing of CRPC treatment, which now includes numerous possibilities based on the six validated drugs, without accounting for those currently under investigation in the ongoing randomized controlled trials. As a consequence, only rational sequencing, which consists in choosing an agent that is not expected to have cross-resistance with previous therapy, can be currently advised.