We sought to develop criteria for ERBB2-positivity (HER2) in colorectal cancer to ensure accurate identification of ERBB2-amplified metastatic colorectal cancer patients suitable for enrollment in a phase II trial of ERBB2-targeted therapy (HERACLES trial). A two-step approach was used. In step 1, a consensus panel of pathologists adapted existing protocols for use in colorectal cancer to test ERBB2 expression and amplification. Collegial revision of an archival test cohort of colorectal cancer samples led to specific recommendations for adapting current breast and gastric cancer criteria for scoring ERBB2 in colorectal cancer. In step 2, from September 2012 to January 2015, colorectal-specific ERBB2 testing protocols and ERBB2 scoring criteria were used to centrally screen for ERBB2-positive KRAS wild-type colorectal cancer patients to be enrolled in the HERACLES trial (clinical validation cohort). In both archival test (N=256) and clinical validation (N=830) cohorts, a clinically sizeable 5% fraction of KRAS wild-type colorectal cancer patients was found to be ERBB2-positive according to the colorectal cancer-specific ERBB2 scoring criteria. ERBB2-positive tumors showed ERBB2 immunostaining consisting of intense membranous ERBB2 protein expression, corresponding to homogenous ERBB2 amplification, in >50% of cells. None of the immunohistochemistry 0 or 1+ cases was amplified. Concordance between SISH and FISH was 100%. In conclusion, we propose specific criteria for defining ERBB2-positivity in colorectal cancer (HERACLES Diagnostic Criteria). In a phase II trial of trastuzumab and lapatinib in a cetuximab-resistant population, HERACLES Diagnostic Criteria shaped the selection of patients and defined ERBB2 as a predictive marker for response to ERBB2-targeted therapy in metastatic colorectal cancer.