To remain competitive, companies must face the changes occurring in their environment and adapt their business processes accordingly. Those processes are implemented in business process management systems (BPMS), which mostly support manual adaptations. That means that the process users have to detect what changes in the environment require process adaptation, and what adaptation operations have to be performed. Such manual adaptations of processes are costly, time-consuming and error prone tasks. For this reason, some contributions of the literature have tried to address the issue of self-adaptations of processes. But these contributions suffer from shortcomings: isolated use of adaptation techniques, non-coverage of the process dimensions and of the adaptation types, etc.; the adaptation issue remains partially addressed. Thus we recommend in this paper a hybrid approach to ensure autonomic adaptations of running processes. According to this approach, the Plan component tries to find an appropriate model version of the concerned process. Then, if such a version does not exist, it looks to reuse an adaptation case that was applied in the past under a similar situation (context). Finally, if necessary, it applies rules, as an artificial intelligence planning technique, to define an ad hoc adaptation. Moreover, the recommended approach takes advantage of the IBM MAPE-K (Monitor, Analyze, Plan, Execute—Knowledge) control loop from autonomic computing, recognized as a prominent solution for self-adaptation at run-time. More precisely the paper addresses the resolution of adaptation needs while covering three process dimensions and all adaptation types and ensuring the separation of concerns for better portability and wide usability through the BPMN standard. It presents both the required Knowledge and the Plan component of the control loop for this resolution. It also shows the effectiveness of the approach by illustrating self-adaptation of a process from the crisis domain, and demonstrates its feasibility by reporting about its implementation and qualitative and quantitative evaluation.