Over the last few years, our view of cellular organization has changed from one in which enzymes and proteins usually act independently to the situation at present where we commonly accept that many, if not all, enzymes act in close association with others. Co-precipitation using an antibody against a test protein is the standard assay for the identification of members of protein complexes [Musso, Zhang and Emili (2007) Chem. Rev. 107, 3585-3600]. The introduction of TAP (tandem affinity purification) tagging enhanced original approaches in order to analyse protein complexes on a larger scale with reduced false discoveries of interacting partners due to more efficient purification of complexes. However, this technique has some limitations as a high-throughput tool for systems biology: the requirement for genetic manipulation to express the tagged protein excludes studies of non-transformable organisms and intact tissue. In those cases where TAP is applicable, a considerable amount of work is required to generate the baits and to optimize experimental conditions. A technique developed in our laboratories, ProCoDeS (Proteomic Complex Detection using Sedimentation), focuses on the detection of endogenous complexes. Protein samples are separated by centrifugation and then different fractions from the resulting gradient are analysed using quantitative MS. The identification of possible protein partners is based on statistical analysis of the co-fractionation of proteins, without any need for purification of individual complexes. The prospects of ProCoDeS and similar techniques based on quantitative MS for measurement of protein complex composition are reviewed in the present article.