Processes have increasingly become an important design principle for complex intra- and inter-organizational e-services. In particular, processes allow to provide value-added services by seamlessly combining existing e-services into a coherent whole, even across corporate boundaries. Process management approaches support the definition and the execution of predefined processes as distributed applications. They ensure that execution guarantees are observed even in the presence of failures and concurrency. The implementation of a process management execution environment is a challenging task in several aspects. First, the processes to be executed are not necessarily static and follow a prede- fined pattern but must be generated dynamically (e.g., choosing the best offer in a pre-sales interaction). Second, deferring the execution of some application services in case of overload or unavailability is often not acceptable and must be avoided by exploiting replicated services or even by automatically adding such services, and by monitoring and balancing the load. Third, in order to avoid a bottleneck at the process coordinator level, a centralized implementation must be avoided as much as possible. Hence, a framework is needed which supports both the modularization of the process coordinator’s functionality and the flexibility needed for dynamically generating and adopting processes. In this paper we show how publish/subscribe techniques can be used for the implementation of process management. We show how the overall architecture looks like when using a computer cluster and publish/subscribe components as the basic infrastructure to drive the enactment of processes. In particular we describe how load balancing, process navigation, failure handling, and process monitoring is supported with minimal intervention of a centralized coordinator.