The standard phase II trial design yields disparate results with similar regimens owing, in part, to variable patient populations enrolled, and appears to be a poor indicator of efficacy demonstrable in a phase III trial. While other phase II trial designs attempt to rectify this problem including the randomized phase II trial and randomized discontinuation trial, they demand more resources and a larger number of patients and are not definitive. A paradigm of multicenter straight phase II trials with rigorous attention to patient selection criteria to improve the validity and reproducibility of results is proposed. Such trials may be superior guides to select regimens for further development than trials that are single-center based with favorable populations.