Pervasive environments are characterized by limited computing resources and wireless connectivity. In parallel, current application domains have variable transactional requirements that do not fit the traditional ACID model. As a result, the pervasive environment characteristics are compelling and cannot be supported by conventional solutions that are typically dedicated to a specific application domain and support a limited set of context parameters. This article aims at providing a complete solution that addresses the challenges of the adaptability of context-aware transactional services “CATS” in pervasive environments. Thus, a new framework CATS-CAE was designed, which offers a comprehensive structure of multiple component chains. The adaptation strategy in CATS-CAE is based on a hybrid approach combining the use of adaptation policies, alternative strategy and behavioral adaptation of composite services through the “Profiled Task Class” concept. A probabilistic model is also presented to support the efficiency of the proposed approach.