Dermal fibroblast cultures from three siblings with a severe form of osteogenesis imperfecta were established in order to analyze their procollagen and collagen synthesis. Cell strains from clinically normal consanguineous parents (first cousins), were also obtained for comparison. Total collagen production in culture media was diminished by 55% in the patients fibroblasts and to a lesser extent in the parents. This decrease was specific for collagenous proteins. From polyacrylamide gel electrophoresis, it appeared that the three children had not only the same defective secretion of pro α 1(I) molecules but that their pro α 1(I) migrated slightly faster than the parental and control counterparts. Analysis of secretion confirmed a reduced rate in procollagen synthesis and the absence of intracellular storage. Upon pepsin treatment, extracellular α 1(I) and α 2(I) chains were found in the expected ratio of 2:1 and migrated normally, suggesting that the altered mobility of pro α 1(I) chains was related to COOH or NH 2 terminal propeptides. In agreement with the reduced type I collagen production, an increase in the α 1( III) α 1( I) ratio was also detected. Furthermore, after a 2.5-h labelling followed by alkylation with iodoacetamide, free intracellular pro α 2(I) and α 1(I) chains were detected in the absence of reduction, consistent with an abnormal intracellular ratio of pro α 1( I) pro α 2( I) that was measured after dithiothreitol reduction. Analysis of intracellular collagen chains from parental strains following a 4-h incubation demonstrated that pro α 1(I) appeared as a doublet, one band with normal mobility and a less intense band migrating faster and corresponding to the defective chain found in the patients. Absence of the abnormal molecules in culture media was related to the demonstration of a defective collagen secretion by parental fibroblasts. Correlation between these biochemical findings and clinical data strongly support a recessive inheritance of the disease that could be classified as a type III form of osteogenesis imperfecta. Patients would be homozygous for the same defective allele and the asymptomatic parents would most likely be heterozygous carriers of the mutation. Although the exact location of the alteration is not yet elucidated, a splicing mutation is suggested.