Women with single, viable pregnancies at 10 + 5 to 13 + 6 weeks, gestation who requested fetal karyotyping for maternal age, parental anxiety, or a previous history of chromosomal aberration were offered participation in this study. With a transabdominal ultrasound-guided technique, early amniocentesis (EA) was performed on 147 women and chorionic villus sampling (CVS) on 174. Spontaneous fetal loss occurred in 6.8 per cent in the EA group and 1.7 per cent in the CVS group. This difference was significant with a confidence interval (CI) of 0.6-9.6 per cent. There was also a significant difference in the need for repeat testing between the groups. In the EA group a repeat test was required in 19.0 per cent due to culture and sample failures, while 5.2 per cent of the women in the CVS group needed repeat testing because of ambiguous results. This prospective study comparing EA and CVS shows that the risk of fetal loss is higher and repeat testing is needed more after EA.