The availability of completed genome sequences of several eukaryotic and prokaryotic species has shifted the focus towards the identification and characterization of all gene products that are expressed in a given organism. In order to cope with the huge amounts of data that have been provided by large-scale sequencing projects, high-throughout methodologies also need to be applied in the emerging field of proteomics. In this review, we discuss methods that have been recently developed in order to characterize protein interactions and their functional relevance on a large scale. We then focus on those methodologies that are suitable for the identification and characterization of protein-protein interactions, namely the yeast two-hybrid system and related methods. Several recent studies have demonstrated the power of automated approaches involving the yeast two-hybrid system in building so-called "interaction networks", which hold the promise of identifying the entirety of all interactions that take place between proteins expressed in a certain cell type or organism. We compare the yeast two-hybrid system with several other screening methods that have been developed to investigate interactions between proteins that are not amenable to conventional yeast two-hybrid screenings, such as transcriptional activators and integral membrane proteins. The eventual adaptation of such methods to a high-throughput format and their use in combination with automated yeast two-hybrid screenings will help in elucidating protein-protein interactions on a scale that would have been unthinkable just a few years ago.