PURPOSE The agent-oriented paradigm has emerged as a viable approach for the development of autonomic systems in the healthcare domain. This paper reviews representative works in this area in order to identify the main research lines and study their level of applicability. Moreover, from the analysis of those works and the authors' own experiences, some lines of future research are suggested. METHODS This paper presents a review of the literature (2002-2008) of applications of agents in healthcare collected from medical databases as well as international conferences. RESULTS 15 agent-based systems were selected according to a set of criteria of work relevancy. Their internal architecture as well as communication-based coordination techniques were reviewed. A classification considering the main goal of the systems has been made. For each category, a discussion of the suitability of agent technology in resolving healthcare problems is presented. CONCLUSIONS Agent-based systems offer added values over classical software approaches (e.g., reusability, reliability, flexibility, robustness, maintainability and adaptability). Furthermore, agent technology supports the integration of legacy systems, tackling the shortcomings of centralised systems, such as performance bottlenecks, resource limitations, and different kinds of failures. During the review, we found a huge number of contributions in this area, showing a growing interest of researchers. However, most of the analysed systems are only prototypes, they are not widely deployed in real environments and they are difficult to extend to others domains. Several aspects (security, reusability, modularity, personalisation, system maintenance) should be studied carefully in the next years in order to bring agent technology to real settings.