The invasion of normal tissues and penetration of basement membranes by malignant cells is likely to require the active participation of hydrolytic enzymes. The four major groups of connective tissue proteins, glycoproteins, proteoglycans, collagen and elastin, vary in their quantitative distributions between different tissues. With the exception of elastin, they also vary qualitatively within each class, so that there are no 'typical' connective tissue barriers to tumor cell penetration. The matrix constituents are stabilized and organized by a variety of covalent and noncovalent interactions between the connective tissue proteins. These interactions play important roles in matrix integrity and may alter the susceptibilities of the constituents to degradative enzymes. It is likely that the complete degradation of the matrix will require the action of more than one enzyme because of differing susceptibilities to tissue proteinases. Primary and transplantable tumors produce well-characterized enzymes which may participate in invasion. These enzymes may also be involved in connective tissue turnover in other normal and pathological situations. The use of long-term tumor cell cultures has verified that tumor cells themselves are capable of producing these enzymes. However, there are many potential modulating influences operative in vivo which are absent in culture so that details of actual mechanisms and control of digestion of complex substrates are not well understood. Recent work on the degradation by tumor cells of extracellular matrices previously produced by cultured cells is likely to shed more light on pathways of tissue destruction in vivo. Experiments with tumor cell variants of defined metastatic potentials will also be useful, but invasive and metastatic abilities are not necessarily correlated. It is unlikely that simple correlations can be drawn between the production of one particular degradative enzyme by all tumor cells and the complex biological mechanisms operative during tumor invasion.