In contrast to an increasing number of agent-based applica- tions in various domains, there has been very little work on maintenance and evolution of agent systems. This paper ad- dresses this gap with a focus on change impact analysis, i.e. estimating the potential eects of changes before they are made as an agent system evolves. We propose a technique for performing impact analysis in an agent system using dy- namic information about agent behaviour. Our approach builds a representation of an agent's behaviour by analyzing its execution traces which consist of goals and plans, and uses this representation to estimate impacts.