BACKGROUND Chorionic-villus sampling is done in early pregnancy to obtain fetal cells for the prenatal diagnosis of genetic and chromosomal defects. Transcervical chorionic-villus sampling has been shown to be safe and effective in national trials. Recently, an alternative transabdominal technique has been suggested as potentially easier and safer. METHODS From April 1987 through September 1989, we prospectively compared transcervical and transabdominal chorionic-villus sampling in 3999 women with singleton pregnancies in whom the risk of a genetically abnormal fetus was increased. Women between 7 and 12 weeks of gestation underwent ultrasonographic evaluation of placental and uterine position. Those with active vaginal infections, active bleeding, or cervical polyps were excluded. If the obstetrician thought either sampling procedure was acceptable, the woman was asked to consent to random assignment to one of the two procedures. Both groups were followed to determine the outcome of pregnancy and the rate of spontaneous fetal loss after chorionic-villus sampling. RESULTS Among the 3999 women who entered the study, sampling was attempted in 3873 (97 percent), 1944 of whom had been assigned to undergo transcervical sampling and 1929 to undergo transabdominal sampling. Of these 3873 women, sampling was eventually successful in 3863. Sampling was successful after a single insertion of the sampling instrument in 94 percent of the transabdominal procedures and 90 percent of the transcervical procedures. Among the women with cytogenetically normal pregnancies who had sampling because of maternal age, the rate of spontaneous fetal loss through 28 weeks of pregnancy was 2.5 percent in the transcervical-sampling group and 2.3 percent in the transabdominal-sampling group (difference, 0.26 percent; 95 percent confidence interval, -0.5 to 1.0 percent). CONCLUSIONS Transabdominal and transcervical chorionic-villus sampling appear to be equally safe procedures for first-trimester diagnosis of fetal abnormalities.