Background: ER stress, triggered by the disruption of intracellular protein homeostasis, promotes the activation of a highly conserved adaptive program called unfolded protein response (UPR). UPR signaling has been reported to contribute to tumor initiation and progression, as well as microenvironment remodeling and chemoresistance in multiple cancer types, including CRC. Several studies point to a direct role of the UPR in tumor angiogenesis and an interaction with EGFR signaling. We therefore hypothesized that genetic variants in ER stress response genes may predict first-line treatment outcome in mCRC pts. Methods: The impact on outcome of 17 functional SNPs in 8 core genes of the ER stress response pathway (IRE1, PERK, ATF6, XBP1, CHOP, GRP78, GADD34, ATF4) was analyzed on a total of 560 pts enrolled in two independent randomized first-line trials: MAVERICC (FOLFIRI/bevacizumab, [bev] n = 163; FOLFOX6/bev, n = 161), and FIRE-3 (FOLFIRI/bev, n = 107; FOLFIRI/cetuximab [cet], n = 129). Genomic DNA from blood samples of pts was genotyped through the OncoArray, a custom array manufactured by Illumina. The association between SNPs and clinical outcomes was evaluated using Cox regression and log-rank tests. The SNP*treatment interaction was assessed in both trials. Results:IRE1 rs16947383 any T allele variant was associated with shorter progression free survival (PFS) and overall survival (OS) compared to the C/C genotype in pts treated with FOLFIRI/bev in the MAVERICC trial both in uni- and multivariate analysis: 10.5 vs 14.1 months for PFS, adjusted hazard ratio (HR) 1.83, 95% confidence interval (CI) 1.12-2.99, adjusted P-value = 0.02; 23.8 vs 38.4 months for OS, adjusted HR 2.12, 95%CI 1.1-4.09, adjusted P-value = 0.02. The association with OS was validated in the FIRE-3 FOLFIRI/bev arm where any T allele carriers showed significantly shorter OS compared to the C/C genotype in both uni- and multivariate analysis: 21.1 vs 44.2 months, adjusted HR 2.06, 95%CI 1.01-4.18, adjusted P-value = 0.05. This effect was not observed in MAVERICC FOLFOX6 and FIRE-3 cet cohorts. A significant interaction was found between IRE1 rs16947383 and treatment on OS in both trials:P = 0.04 in MAVERICC (FOLFIRI vs FOLFOX6), and P = 0.03 in FIRE-3 (bev vs cet). In the MAVERICC FOLFOX6/bev arm, XBP1 rs2239815 and PERK rs6731022 SNPs were associated with shorter PFS and OS, while GADD34 rs610308 with longer OS, in multivariable analysis (P< 0.05). Conclusions: Our results provide evidence that germline polymorphisms in ER stress response genes may have a predictive value in mCRC pts receiving first-line bev-based treatment and contribute to modulate anti-VEGF efficacy.