To compare the classification CAPRA (based on clinical-pathological findings) and minimal residual disease (MRD) (based on biological characteristics) to predict biochemical failure (BF).The clinical-pathological findings of the prostate biopsy were used to determine the CAPRA score, classifying patients into low, intermediate and high risk. Blood and bone marrow samples to detect circulating prostate cells (CPCs) and micro-metastasis were taken. The samples were classified as positive if ≥1 prostate cell was detected, forming three subgroups; Group A (MRD negative), Group B (micro-metastasis positive, CPC negative) and Group C (CPC positive). Patients were followed-up for 10 yearsor BF. Kaplan-Meier biochemical failure free survival (BFFS) curves, a predictive flexible parameter survival model and mean restricted survival times (MRST) were determined.347 men participated, BF risk increased with increasing CAPRA score, HR 1.21 intermediate, 1.64 high risk; versus MRD HR 1.91 and 4.43 for Groups Band C. After 10 years the BFFS and MRST were 76%, 50% and 17% and 9, 7 and 5 years respectively for CAPRA versus 94%, 57% and 26% and 10, 9 and 6 years respectively for MRD. The concordance between observed and predicted BFFS was acceptable for CAPRA (Harrell´s C 0.64) and very good (0.92) for MRD. The BFFS curves for MRD were not proportional with time, they were similar for 5 years for Groups A and B, with increasing BFFS in Group B there after.The CAPRA score did not distinguish between Groups A and B, one third of low risk CAPRA patients had CPCs detected.The MRD classification was superior to CAPRA, differentiating between early and late failure.