A biomarker as defined by a National Institutes of Health workshop in 2001 is "a characteristic that is objectively measured as an indicator of normal biologic or pathologic processes or pharmacologic responses to a therapeutic intervention." But all biomarkers serve the same function, and there are 2 different classes of biomarkers. The biomarker class I wish to discuss is the one that can be accepted from a regulatory perspective. Otherwise, when we talk about biomarkers in general, they can be almost any scientific data extracted from biologic phenomena. From the National Cancer Institute's standpoint, there are important classification distinctions, developed in concert with the US Food and Drug Administration, in which biomarkers can be defined as either prognostic or predictive. As such, a distinction can be made between a "biomarker" and a "qualified biomarker." Even so, the ideal form of a biomarker is when it acts as a "surrogate biomarker." A surrogate biomarker is a marker that can be reliably used as a substitute for a clinical endpoint in clinical trial circumstances. In cancer clinical trials when they are conducted to prove clinical efficacy, the key endpoint in clinical trials is primarily survival. Acceptable surrogate biomarkers thus are few in number because it is a requirement that they must show an incontrovertible linkage between the biomarker and the clinical end result such as prediction of extension of survival.