Background: In a previous work we demonstrated that pts harboringTYMS-3’UTR+6 bp/+6 bp and ERCC1-118 C/T or C/C genotypes may better receive capecitabine instead of 5FU in an OXA-based first-line treatment. The objective of the present work was to demonstrate that prospective selection of first-line treatment according to these genetic variants could improve the outcome of metastatic CRC pts. Methods: We designed a phase II comparative clinical trial (SETICC, SElected Treatment In Colorectal Cancer; NCT01071655) in which 195 CRC pts with initially unresectable metastasis were randomized 2:1 to receive XELOX plus bevacizumab in a control arm or, based on our previous results, chemotherapy according to their TYMS-3’UTR and ERCC1-118 genotypes plus bevacizumab in in the experimental arm (table 1). Results: In 161 evaluable pts the overall best objective response (Complete +Partial response) was 47% (95 CI: 33.9-61.1) in the control arm and 65% (95 CI: 55.2-73.9) in the experimental arm (p = 0.04). In the Per Protocol (PP) population (N = 180), 20% (N = 37) were converted to resectable, 16 pts in the control arm and 21 in the experimental arm. The R0 surgery was 43% (N = 7) in the control arm and 86% (N = 18) in the experimental arm (p = 0.018). Median PFS (10 months; 95 CI: 8.5-11.6 vs 9.4 months; 95% CI: 6.7-12.1 months) and overall survival (19.3 months; 95% CI: 15.5-23.1 vs. 16.5 months; 95% CI: 13.5-19.5.) was better for pts in the experimental arm (p:NS). Conclusions: This study shows for the first time that prospective selection of first-line treatment according to TYMS-3’UTR and ERCC1-118 polymorphisms significantly improves the ORR and R0 liver metastasis resection of CRC pts with non-initially resectable liver metastases. This approach should be taken into account in this kind of pts in order to increase the probability of R0 surgery success rate. Clinical trial information: NCT01071655