Background: miR 31 3p expression has previously shown correlation with outcomes in KRAS wild-type (wt) aCRC patients receiving EGFR-targeted therapy. We have therefore evaluated miR 31 3p in a large randomized trial of panitumumab. The a priori hypothesis was that pts with the lowest miR 31 3p expression would have better outcomes and increased benefit from panitumumab. Methods: miR 31 3p was measured in tumor from 213 pts randomized to irinotecan (Ir, n = 111) or irinotecan/panitumumab (IrPan, n = 102) in a trial of second line therapy for aCRC (PICCOLO). The analysis population comprised RAS wt (KRAS and NRAS wt) pts (n = 188: Ir = 101, IrPan-87). End-points were progression-free survival (PFS), overall survival (OS), complete/partial response rate (RR) and disease control rate (DCR). The predefined model divided pts into 3 tertiles with high, intermediate (int) and low miR 31 3p expression, and compared outcomes and treatment effects across these groups. Multivariate analysis was performed, adjusting for Köhne score. Results: In the prognostic analysis, compared with low expression pts and after adjustment for treatment arm, int and high pts had worse OS (HR 1.58, 2.03 respectively; p = 0.0012) and worse PFS (HR 1.60, 1.60 respectively; p = 0.018). In multivariate analysis, miR 31 3p and Köhne score were independently associated with OS (p = 0.0006 and p = 0.002 respectively). miR 31 3p was also significantly independently associated with RR (p = 0.015) and DCR (p = 0.074). In the predictive analysis, panitumumab produced marked PFS benefit in pts with low and int miR 31 3p expression (HR = 0.50 [p = 0.019] and HR = 0.57 [p = 0.031] respectively), but not in pts with high expression (HR 0.72, p = 0.23); however, a statistically significant treatment/expression interaction was not seen. Conclusions: Pts with lower miR 31 3p have significantly better OS, PFS, RR and DCR, independent of treatment. Pts with low/int miR 31 3p had significant PFS benefit from panitumumab whilst pts with high miR 31 3p did not; however this study was not powered to demonstrate a statistically significant treatment/expression interaction. miR-31-3p is a highly promising biomarker in aCRC.