Collagens are a family of extracellular matrix proteins that are critically important for providing structural support to tissues and for regulating cell behaviour. The thesis describes the biochemical analysis of collagen binding by two globular proteins that profoundly influence collagen fibril formation in vivo, decorin and SPARC. Decorin is the archetypal small leucine-rich repeat proteoglycan and an important regulator of collagen fibrillogenesis. The crystal structure of decorin, published in 2004, revealed a dimeric structure in which the presumed collagen binding site was not accessible. Whether the dimer is functional has been controversial, however, and it has been claimed that biologically active decorin is monomeric. In order to resolve this controversy, I designed a number of decorin mutants to disrupt the crystallographic dimer, including two mutants which introduced glycosylation sites into the dimer interface. Size exclusion chromatography with multi-angle laser light scattering and analytical ultracentrifugation were used to determine the oligomeric states of decorin and the designed mutants. I found that wild-type decorin dimerises in solution with a dissociation constant of ~1 μM. The mutants with engineered glycosylation sites were pure monomers while other mutants remained dimeric. Thermal unfolding experiments showed that the engineered decorin monomers were as stable as wild-type decorin. Mutations on the concave face of decorin abolished collagen binding, regardless of whether the mutant proteins retained the ability to dimerise or not. Thus the concave face of decorin is involved in collagen binding and the dimer must dissociate in order to bind collagen. The crystal structure of human SPARC bound to a collagen-like triple-helical peptide was determined in the Hohenester lab in 2008. The key collagen binding residues are conserved between human and invertebrate (Drosophila and C. elegans) SPARCs. There is a key difference between the orthologues, however: high-affinity collagen binding to human SPARC requires proteolytic cleavage of an inhibitory loop that is absent from invertebrate SPARCs. To investigate the functional consequences of this structural difference for collagen binding, I compared the interactions of the different SPARCs with collagen I and IV, using solid-phase assays and surface plasmon resonance. I found that invertebrate SPARCs did not bind collagen more tightly than human SPARC, suggesting that the absence of the inhibitory loop does not confer a higher affinity for collagen in invertebrate SPARCs. I made several unsuccessful attempts to crystallise an invertebrate SPARC, including experiments in which the glycan was trimmed by endoglycosidase digestion or removed by mutagenesis of acceptor sites.