This paper examines the role of the extracellular matrix (ECM) in the development of the cornea. After a brief summary of the corneal structure and ECM, we describe evidence suggesting that the differentiation of neural crest (NC) cells into endothelium and fibroblasts is under the control of ocular ECM. We then examine the role of collagen I in stromal morphogenesis by comparing normal corneas with those of homozygous Mov 13 mice which do not make collagen I. We report that, in spite of this absence, the cellular morphology of the Mov13 eye is indistinguishable from that of the wild type. In the 16-day mutant stroma, however, the remaining collagens form small amounts of disorganized, thin fibrils rather than orthogonally organized 20 nm-diameter fibrils; a result implying that collagen I plays only a structural role and that its absence is not compensated for. It also suggests that, because these remaining collagens will not form the normal fibrils that they will in vitro, fibrillogenesis in the corneal stroma differs from that elsewhere. The latter part of the paper describes our current work on chick stromal deposition using corneal epithelia isolated with an intact basal lamina that lay down in vitro approximately 3 microns-thick stromas of organized fibrils similar to that seen in vivo. This experimental system has yielded two unexpected results. First, the amount of collagen and proteoglycans produced by such epithelia is not dependent on whether its substratum is collagenous and we therefore conclude that stromal production by the intact epithelium is more autonomous than hitherto thought. Second, chondroitin sulphate (CS), the predominant proteoglycan, appears to play no role in stromal morphogenesis: epithelia cultured in testicular hyaluronidase, which degrades CS, lay down stromas whose organization and fibril-diameter distribution are indistinguishable from controls. One possible role for CS, however, is as a lubricant which facilitates corneal growth: it could allow fibrils to move over one another without deforming their orthogonal organization. Finally, we have examined the processes of fibrillogenesis in the corneal stroma and conclude that they are different from those elsewhere in the embryo and in vitro, perhaps because there is in the primary stroma an unidentified, highly hydrated ECM macromolecule that embeds the fibrils and that may mediate their morphogenesis.