During reorganization of collagen gels by human skin fibroblasts the total protein content of the gels remained approximately constant. Only 5% of the collagen was degraded, although the volume of the gels decreased by 85% or more. It could be concluded, therefore, that gel reorganization required physical rearrangement of pre-existing collagen fibrils rather than degradation of the original collagen and resynthesis of a new matrix. Collagen molecules in the gels were not covalently crosslinked or otherwise modified enzymically during gel reorganization, as determined by sodium dodecyl sulphate/polyacrylamide gel electrophoresis and collagen repolymerization studies. Serum was required for gel reorganization and, in the absence of serum, cell spreading was predominantly filipodial, i.e. there was little cytoplasmic reorganization. At the electron-microscopic level it was found that many more collagen fibrils became associated with the cells in the presence of serum than in its absence. Serum was also found to promote the synthesis and secretion of proteins by the cells, and conditioned medium could take the place of serum in promoting gel reorganization. The involvement of cell-secreted factors was also demonstrated by the ability of cycloheximide to inhibit gel reorganization. Finally, when gel reorganization was stopped by adding cytochalasin D to the incubations or removing cells by detergent treatment, a small but significant re-expansion of the collagen fibrils was observed. Consequently, a portion of the collagen that had been physically reorganized by the gels was unstable and could not hold its position without continued force exerted by the cells.