We present a generic architecture for content based retrieval of images, which can be extended to the requirements of large distributed and heterogeneous collections. The system is modeled as a multi agent system where an autonomous search agent encapsulates independent image retrieval algorithms. An optimal team of agents is dynamically selected for every retrieval problem. A flexible protocol allows for dynamic addition of search agents incorporating new pattern recognition techniques. These agents are coded as mobile agents, so that they can travel across a wide area network and analyze the documents at their sources. The separation of physical image forms and their logical structural composition allows the search agents to operate over heterogeneous repositories. A prototype implementation validates the effectiveness of the architecture. © 1999 IEEE.