We propose a methodology for conducting a phase II clinical trials in a setting where the disease is categorized into multiple subtypes and the primary outcome is binary. The hierarchical model, which is tailored to each particular application, allows treatment effects to differ across subtypes while assuming a priori that the effects are exchangeable and correlated. We describe an application to a trial for sarcoma in which the treatment activity is characterized by tumor response. The application illustrates how the hierarchical Bayesian model borrows across subtypes. In addition, we present several illustrative examples to demonstrate how the methodology differs from other approaches. A demonstration of the software used to simulate this design is presented.