A criterion has been devised for the assessment of intermolecular contacts between protein subunits coupled to a gel matrix. After reversible coupling by way of disulfide groups to Sepharose 4B, the system is exposed to a bifunctional reagent. The protein is then released by reduction, and the proportion of dimers and higher oligomers formed by cross-linking is determined by gel electrophoresis, followed by densitometry of the stained gels. Experiments were performed with G-actin coupled to dithio-2-dipyridyl Sepharose 4B. At low concentration of coupled protein, no species other than the monomer were obtained; under these conditions any intermolecularly cross-linked species would represent pre-existing associated states coupled to the matrix. At higher protein concentrations dimers and higher species progressively appear. Their proportion is much higher than predicted on the basis of a random statistical distribution of coupled molecules throughout the accessible internal volume of the matrix. It follows that protein-protein contacts can occur either because of high flexibility of the polysaccharide chains of the matrix, or because the protein is partly (but not wholly) present as a shell on the surface of the beads. With Affi-gel 10, which has N-hydroxysuccinimide ester coupling groups, similar experiments were performed, taking advantage of the degradation of the matrix under relatively mild acid conditions. In this case the degree of cross-linking was much lower than in the Sepharose 4B system at the same protein concentration. However, this medium proved unsatisfactory for the measurement of interactions of the bound actin with other muscle proteins present in the mobile phase. The results with Sepharose 4B support the validity of previous studies on interaction of monomeric actin with other muscle proteins.