Background: The phase II randomized AtezoTRIBE study demonstrated that the addition of atezolizumab (atezo) to first-line FOLFOXIRI plus bevacizumab (bev) prolongs PFS of mCRC patients, but benefit is limited among patients with proficient mismatch repair (pMMR) tumours. Among these patients, identifying a subgroup able to achieve benefit from immune-checkpoint inhibitors is a crucial challenge of translational research. To this end, we investigated the potential predictive role of an immunomodulatory gene expression signature (IO score), measuring the presence of infiltrating inflammatory cells and differentiated stromal microenvironment. Methods: AtezoTRIBE was a phase II comparative trial in which mCRC patients, unselected for MMR status, were randomized 1:2 to receive first-line FOLFOXIRI/bev (control arm) or FOLFOXIRI/bev/atezo (experimental arm). RNA was obtained from FFPE blocks of tumour specimens collected at baseline from 142 (65%) out of 218 enrolled patients. RT-qPCR was performed using DetermaIO™, to assess mRNA expression of a targeted panel of 27 genes. The established pan-cancer IO score and threshold were applied to dichotomize tumours as IO-positive (IO⁺) or IO-negative (IO^-). Results: IO score was successfully determined in 122 (86%) cases, and 33 tumours were defined as IO⁺ (27%). No differences in terms of baseline clinical and molecular features were observed between IO⁺ and IO^- tumours. Patients with IO⁺ and IO^- tumours showed similar PFS (median PFS: 14.4 vs 13.6; HR 0.84 [95%CI: 0.53-1.33], p = 0.468). An interaction between IO status and treatment effect was reported (p for interaction = 0.066), with higher PFS benefit in favour of the experimental arm among patients with IO⁺ tumours (HR 0.39 [95% CI:0.15-1.02]) than among those with IO^- tumours (HR 0.83 [95% CI 0.50-1.35]). A similar trend was observed in the pMMR subgroup (n = 110) (IO⁺ tumours: HR for PFS 0.47 [95% CI 0.18-1.25]; IO^- tumours: HR for PFS 0.93 [95% CI 0.56-1.55]) (p for interaction = 0.139). No differences in terms of ORR were reported between arms according to the IO status. Conclusions: The investigated immunomodulatory signature (IO score) may be helpful to predict benefit from the addition of atezolizumab to first-line FOLFOXIRI/bev in metastatic colorectal cancer, also in the cohort of pMMR tumours. Our results support the hypothesis that a deeper characterization of tumour immune microenvironment may help identifying mCRC patients more likely to benefit from ICI-based therapeutic strategies. These findings are worthy of further investigation in independent cohorts.