Multi-agent systems (MASs) play an important role in information technology. They are used in many applications. To develop these systems requires the developer to deal with several issues and to implement many of the system components, such as protocols and agents' functionalities. Due to their significant role, agent toolkits are becoming more of a necessity to help build, re-use, and deploy agents. Although there are several agent-building toolkits that can help developers to effectively develop MAS and allow them to focus more on the application to specific domains, it is difficult for developers to select the appropriate toolkit. This paper presents a methodology to evaluate MAS toolkits based on the criteria of their availability, environment, development, characteristic properties and performance.