Summary: The prediction of protein–protein complexes from the structures of unbound components is a challenging and powerful strategy to decipher the mechanism of many essential biological processes. We present a user-friendly protein–protein docking server based on an improved version of FRODOCK that includes a complementary knowledge-based potential. The web interface provides a very effective tool to explore and select protein–protein models and interactively screen them against experimental distance constraints. The competitive success rates and efficiency achieved allow the retrieval of reliable potential protein–protein binding conformations that can be further refined with more computationally demanding strategies.

Availability and Implementation: The server is free and open to all users with no login requirement at http://frodock.chaconlab.org

Contact:  pablo@chaconlab.org

Supplementary information:  Supplementary data are available at Bioinformatics online.