Background: There are no known markers for predicting AA/P efficacy in chemo-naïve mCRPC stage. We explored a model for predicting the 12-week efficacy outcome after initiation of AA/P in a prospective clinical trial (NCT# 01953640). Methods: mCRPC patients initiating AA/P underwent biopsy of metastases at baseline (pre AA/P) and at week 12. Somatic whole exome DNA sequencing; gene expression for AR full length (ARFL) and splice variants (ARVs) was performed of baseline metastases. Composite progression on AA/P at week 12, (primary endpoint) was evaluated with PSA, RECIST, bone scan and symptoms (based on PCWG2 criteria). Candidate molecular factors (RNA expression of ARFL, ARV3, ARV7, ARV9, ARV23, ARV45, Chromogranin-A (CGA)) were explored in a multivariate Cox Proportional Hazard Regression (CPHR) model. Additionally, PSA/testosterone levels, Gleason Score (GS:2-6;7;8-10) at initial diagnosis; time from starting hormone therapy to mCRPC stage (years), serum CGA levels were also included. A final CPHR model fitted only those factors which exceeded an entry threshold in univariate analysis or were considered clinically relevant. Results: Between 6/2013 and 11/2014, 56/70 patients enrolled had disease assessed at the 12-week time point. The CPHR model includes 56 patients with complete sequencing and 12 week clinical outcomes. The median follow up for the cohort was 170 days (41-502 days). Progression at 12-weeks was observed in 23/56 patients. The final CPHR model included baseline expression of ARV45 (HR = 13.107, p-value = 0.017); CGA levels (HR = 0.998, p-value = 0.004); time from hormone therapy to mCRPC stage (HR = 0.782, p-value = 0.008); GS at initial diagnosis (HR = 0.004, p-value < 0.001 for GS group 7 and HR = 0.008 p-value < 0.001 for GS group 8-10) and ARFL, ARV3, ARV7, ARV9, ARV23. A C-statistic for the model was observed at 0.76 with an adjusted R² of 0.47 (likelihood ratio p-value < 0.001). Conclusions: Clinical prediction rules are needed for developing precision medicine in mCRPC. An initial attempt to incorporate molecular and clinico-pathological factors appears feasible. Evaluation for model stability remains ongoing. Clinical trial information: NCT# 01953640.