Development of targeted therapies is accelerating in response to widespread identification of hypothesized biomarkers. Of particular interest are candidate predictive markers believed to be related to the efficacy of an experimental treatment under study, where co-primary aims of a phase II trial may be determination of the marker’s predictive value and identification of the marker-related subpopulation most likely to benefit. Before a new biomarker can be used to guide treatment decisions and patient care, however, a lengthy process from marker identification to validation must occur. For quantitative biomarkers (e.g., circulating levels of a target), a threshold to distinguish marker-low from marker-high patients may additionally be required. This process becomes inefficient when individual steps are accomplished in a post-hoc manner using data from multiple and potentially disparate sources, and results may be biased or confounded if marker identification and marker evaluation studies are performed separately or without a prospective framework.