Background: Previous data suggest the EGF ligands EREG/AREG may predict outcome of KRAS wt patients (pts) in the chemo-refractory setting but has not been previously reported from first line randomised trials. Methods: FFPE samples from primary tumors of pts in Arms A&B of the COIN trial of Ox fluoropyrimidine (Fp) +/- cet were analysed for EGFR IHC, KRAS/NRAS/BRAF mutation and EREG/AREG expression by RT-PCR. Ligand levels were assessed against baseline data, prognostic markers as uni/multivariate analyses and as predictive markers in wild type (wt) and mutant (mt) cohorts and separately by Fp backbone [capecitabine (CapOx) or 5FU (FOLFOX)]. Tests for interaction were performed with EREG/AREG continuous, using Flexible Parametric survival analysis. Optimal cutoffs for predictive effects were defined using point at which 95% CI first excluded zero. Results: 952/1630 (57%) of pts were evaluable for all parameters. EREG/AREG were highly correlated Pearson’s rho (ρ) = 0.74; p<0.0001. High EREG/AREG levels were associated with KRAS wt (p<0.005) and with primary tumor in left colon/rectum, presence of liver metastases and high CEA (p<0.05). In the control arm, high EREG/AREG conferred a better prognosis among KRAS wt pts in multivariate. EREG superseded AREG in a combined model. High EREG predicted for OS benefit in KRAS wt pts treated with FOLFOX +cet, with optimal cut-off 80th centile. OS HR for ±cet ≥80th centile =0.33, 95% CI 0.14-0.78, p=0.011; <80% centile HR=0.99, 95% CI 0.67-1.47, p=0.96; interaction p=0.059; ≥50th centile HR=0.66, 95% CI 0.40-1.09, p=0.11; <50% centile HR=1.09, 95% CI 0.66-1.81, p=0.73; interaction p=0.17. Similar results were seen for PFS, with optimal cut-off 50th centile. There was no predictive effect for pts treated with CapOx. Conclusions: The data suggest a prognostic effect of EREG/AREG in aCRC. The original hypothesis, that KRAS wt patients with high EREG expression have improved outcome with cet, seems to be limited to patients treated with FOLFOX in the first-line setting.