We completed prenatal diagnostic studies from 129 pregnancies at risk for osteogenesis imperfecta (OI). Studies in 107 pregnancies were completed by analysis of collagen synthesized by cells cultured from chorionic villus biopsies and the remaining 22 used direct mutation identification or analysis of polymorphic restriction sites in the COL1A1 gene of type I collagen. The vast majority of studies (n=113) were obtained to identify fetuses with OI type II (the perinatal lethal form) and some fetuses affected with OI type III or IV (the deforming varieties). Of the 50 couples who had had one previous affected pregnancy with the lethal form of OI, one had a second affected pregnancy, a rate of 2 per cent. Two of the seven unaffected couples (28 per cent) who had had two previous affected pregnancies with OI type II had a third affected pregnancy; none of the three with two previous pregnancies with OI type III had a third. Pregnancies at risk for OI type I could not be ascertained reliably by biochemical analysis of cultured CVS cells but were identified by direct analysis of the causative mutation or the use of linked markers in families. All prenatal diagnostic studies were undertaken only after earlier diagnostic studies (biochemical or molecular) had been completed on the proband, a necessary strategy for accurate results. In all pregnancies at risk for OI type II, OI type III, and OI type IV studied with biochemical strategies and in pregnancies at risk for OI type I studied with molecular techniques, there were neither false-negative nor false-positive results. Diagnostic information can be obtained within 20-30 days of biopsy using biochemical techniques and within 10-14 days when molecular strategies are used. 1997 by John Wiley & Sons, Ltd.