GpDB is a publicly accessible, relational database of G-proteins and their interactions with GPCRs and effector molecules. The sequences are classified according to a hierarchy of different classes, families and sub-families, based on extensive literature search. The main innovation besides the classification of both G-proteins and GPCRs is the relational model of the database, describing the known coupling specificity of the GPCRs to their respective alpha subunit of G-proteins and also the interaction between G-protein subfamilies and specific effector types, a unique feature not available in any other database. There is full sequence information with cross-references to publicly available databases, and the user may submit advanced queries for text search. Furthermore there is an interface for running BLAST against the database and a pattern search tool. The database will be very useful for the study of G-protein/GPCR and G-protein/effector interactions and for future development of algorithms predicting this interaction.