The small-molecule epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors and the anti-EGFR monoclonal antibodies have proven activity in lung and colorectal adenocarcinomas, respectively, but only a small fraction of patients exhibit significant responses. The observation that only a minority of patients respond to EGFR-targeted therapies, in combination with their toxicity and high costs, has driven the search for molecular markers predictive of response. The main focus of the present review is the recent discovery that mutations in the KRAS oncogene constitute a negative predictive marker in this clinical setting, namely that their presence can be used to predict which patients are unlikely to benefit from treatment with EGFR-directed therapy.