Predictive biomarkers capable of discriminating individuals who will benefit from a given therapy from those who will not are key to personalized medicine. New drugs are developed in patients with advanced disease, when ethical and practical shortcomings limit the collection of tumor specimens. The neoadjuvant setting offers a unique opportunity for overcoming these limitations. Tumor samples are collected at diagnosis and posttreatment surgery as part of a routine therapeutic approach. Pathological complete response has been consistently associated with long-term survival and may be useful as an intermediate endpoint in developing and evaluating predictive, as well as surrogate, biomarkers. In window-of-opportunity studies, new drugs are administered shortly before planned surgery, and the effect of the intervention can be assessed by comparing diagnostic biopsy with the surgical specimen. Until now, clinically useful and validated predictive markers for targeted therapy are rare, but appropriate investigations in neoadjuvant studies will likely change this.