An accompanying commentary notes the threshold that would lead to the regimens being tested in a conventional phase 3 design was a failure-free survival of 50% at 10 months. This measure was met for both groups in patients who had wild-type KRAS and were even better—as expected—for patients who also had wild-type NRAS, and BRAF. The author suggests these data show the need for proper molecular classification of colorectal cancer, which would enable clinicians to select the most effective treatment for patients. He adds that under the selection pressure of continuous treatment, emerging RAS mutant clones could cause resistance to anti-EGFR antibodies and intermittent treatment might help to prevent the emergence of such clones. He concludes that the findings of this trial cannot be translated into clinical practice without further validation in randomised phase 3 trials, but it does provide a clear hypothesis to test.