Web service technology expands the role of the Web from a simple data carrier to a service provider. To sustain this role, some issues such as reliability continue to hurdle Web services widespread use, and thus need to be addressed. Autonomic computing seems offering solutions to the specific issue of reliability. These solutions let Web services self-heal in response to the errors that are detected and then fixed. Self-healing is simply defined as the capacity of a system to restore itself to a normal state without human intervention. In this paper, we design and implement a self-healing approach to achieve Web services reliability. Two steps are identified in this approach: (1) model a Web service using two behaviors known as operational and control, and (2) monitor the execution of a Web service using a control interface that sits between these two behaviors. This control interface is implemented in compliance with the principles of aspect-oriented programming and case-based reasoning. © 2011 IEEE.