Among patients with the same clinical disease diagnosis, response to a treatment is often quite heterogeneous. For many diseases this may be due to molecular heterogeneity of the disease itself, which may be measured via a biomarker. In this talk, we consider the problem of evaluating clinical trial designs for drug response based on an assay of a predictive biomarker. In the planning stages of a clinical trial, one important consideration is the number of patients needed for the study to be able to detect some clinically meaningful difference between two treatments. We outline several marker validation trial designs in terms of the scientific questions each one is able to address, and compute the number of patients required for each one. We exhibit efficiency graphs for several special cases to summarize our results.