Many real-time systems use preemptive priority-based scheduling in their internals to guarantee certain real-time performance. This includes technologies that range from The Real-Time Specification for Java (RTSJ) to middleware like Real-Time Common Object Request Broker Architecture (RT-CORBA), which offers additional models and policies that blend client and server information. This decision eases the integration of real-time admission tests and dispatching policies in these types of infrastructures. In this paper, we analyze different trade-offs that emerge from the definition of different propagation models for distributed real-time Java. The paper covers technological integration aspects as impact on interfaces and other practical issues mainly related to the performance that this model offers to a real-time application and non-functional overhead. The contribution described in the paper may help in the development of The Distributed Specification for Java (DRTSJ). Copyright © 2011 John Wiley & Sons, Ltd.