Background: Patients with metastatic colorectal cancer whose disease has progressed on oxaliplatin- and irinotecan-containing regimens may benefit from EGFR-inhibiting monoclonal antibodies if they do not contain mutations in the KRAS gene (are "wild type"). It is unknown whether these antibodies, such as cetuximab, are more efficacious in refractory metastatic colorectal cancer as monotherapy, or in combination with irinotecan. Lack of mutation in KRAS, BRAF and PIK3CA predicts response to EFGR-inhibitors. The ICECREAM trial examines the question of monotherapy versus combination with chemotherapy in two groups of patients: those with a "quadruple wild type" tumour genotype (no mutations in KRAS, NRAS, PI3KCA or BRAF genes) and those with the specific KRAS mutation in codon G13D, for whom possibly EGFR-inhibitor efficacy may be equivalent. Methods and design: ICECREAM is a randomised, phase II, open-label, controlled trial comparing the efficacy of cetuximab alone or with irinotecan in patients with "quadruple wild type" or G13D-mutated metastatic colorectal cancer, whose disease has progressed on, or who are intolerant of oxaliplatin- and fluoropyrimidine-based chemotherapy. The primary endpoint is the 6-month progression-free survival benefit of the treatment regimen. Secondary endpoints are response rate, overall survival, and quality of life. The tertiary endpoint is prediction of outcome with further biological markers. International collaboration has facilitated recruitment in this prospective trial of treatment in these infrequently found molecular subsets of colorectal cancer. Discussion: This unique trial will yield prospective information on the efficacy of cetuximab and whether this is further enhanced with chemotherapy in two distinct populations of patients with metastatic colorectal cancer: the "quadruple wild type", which may 'superselect' for tumours sensitive to EGFR-inhibition, and the rare KRAS G13D mutated tumours, which are also postulated to be sensitive to the drug. The focus on establishing both positive and negative predictive factors for the response to targeted therapy is an attempt to improve outcomes, reduce toxicity and contain treatment costs. Tissue and blood will yield a resource for molecular studies. Recruitment, particularly of patients with the rare G13D mutation, will demonstrate the ability for international collaboration to run prospective trials in small colorectal cancer molecular subgroups. Trial registration: Australian and New Zealand Clinical Trials Registry: ACTRN12612000901808 , registered 16 August 2012.