In recent years, the clinical development of targeted therapies has been advanced by a greater understanding of tumor biology and genomics. Nonetheless, drug development remains a slow and costly process. An additional challenge is that targeted therapies may benefit only a subset of patients treatedtypically those patients whose tumors are dependent on the target of interest. Thus, there is a growing need for the incorporation of both predictive and pharmacodynamic (PD) biomarkers in drug development. Predictive biomarkers are important to help guide patient selection, while PD biomarkers can provide information on the pharmacologic effects of a drug on its target. PD studies may provide insights into proof of mechanism (i.e., Does the agent hit its intended target?) and proof of concept (i.e., Does hitting the drug target result in the desired biologic effect?). PD studies may also provide information on the optimal biologic dosing or scheduling of a targeted agent. Herein, we review PD endpoints in the context of targeted drug development in non-small cell lung cancer, highlighting some of the key challenges encountered to date. In doing so, we discuss recent experiences with repeat tumor biopsies, surrogate tissue analysis, alternative clinical trial designs (e.g., window-of-opportunity trials), circulating biomarkers, and mechanism-based toxicity assessments. The application of such technologies and biomarkers in early clinical trials may facilitate rational drug development, while enhancing our understanding of why certain targeted therapies succeed or fail.