Gathering functionally similar agent-based Web services into communities has been proposed and promoted on many occasions. In this paper, we compare the performance of these communities with self-managed, single agent-based Web services from trust perspective. To this end, we deploy a reputation model that ranks communities and Web services with respect to different reputation parameters. By relating the parameters, we extend our discussion to analyze the beneficial cases and incentives for a single Web service to join a community even if this joining could negatively impact other parameters. Besides theoretical discussions of this analysis, we discuss the system implementation along with simulations that depict diverse parameters and system performance. © 2010 IEEE.