Work by a large number of investigators over the last decade has established that over 90% of patients with osteogenesis imperfecta have mutations in one of the two genes for type I procollagen, that most unrelated probands have different mutations in the genes, and that the mutations found in most of the serious variants of the disease cause synthesis of abnormal pro alpha chains of the protein. The results have demonstrated that synthesis of structurally abnormal but partially functional pro alpha chains can interfere with folding of the central region of the protein into a triple-helical conformation, prevent processing of the N-terminal propeptides of procollagen, or produce subtle alterations in conformation that interfere with the self-assembly of the protein into collagen fibrils. One of the unsolved mysteries about the disease is why some mutations produce severe phenotypes, whereas very similar mutations produce mild phenotypes. Recent studies in transgenic mice suggest that nongenetic factors, such as stochastic events during development, may determine the severity of the disease phenotype produced by a specific mutation. Also, recent results raised the possibility that strategies of antisense gene therapy may be effective in treating the disease some time in the future. Specific inhibition of expression of a mutated collagen gene has been obtained with antisense oligonucleotides in cell culture experiments. However, there is no means of selective delivery of antisense oligonucleotides to the appropriate tissues.