The pregnancy outcome of 1936 women who had transcervical chorionic villus sampling (CVS) with a flexible biopsy forceps was evaluated. Follow-up until 4 weeks after delivery was 99.4 per cent. Various patient- and procedure-related risk factors for spontaneous loss (fetal or neonatal death) were analysed using stepwise logistic regression analysis. The overall spontaneous loss rate was 4.5 per cent. Factors found to be significantly associated with spontaneous loss were quantity of villi < or = 15 mg (relative risk (RR) 2.13), a history of first-trimester miscarriage (RR 1.87) or delivery between 16 and 27 weeks (RR 3.87), cervical culture positive for anaerobes (RR 4.52) or group B streptococcus (RR 3.62), post-procedural bleeding > 3 days (RR 1.99), and multiple insertions (RR 2.64). Significant differences in loss rates between individual operators were found. A learning effect was not present. There were no infants born with terminal transversal limb anomalies in our series. We conclude that knowledge about significant risk factors for spontaneous loss after CVS is important both for obstetricians carrying out CVS procedures and for women seeking prenatal diagnosis.