Chorionic villus sampling (CVS) was performed in 10,000 consecutive singleton pregnancies by a single principal operator, working in two institutions. The procedure was performed between 8 and 32 gestational weeks: transabdominal (TA) sampling was carried out in 8479 cases and transcervical (TC) in 1521. Patients were referred for chromosomal risk in 89.1 per cent of cases, Mendelian disorders in 10.5 per cent, and DNA investigations for paternity or infectious agents in 0.4 per cent of cases. The sampling success rate for both TA and TC techniques by the second insertion was 99.8 and 99.2 per cent, respectively. TA sampling succeeded in a higher number of cases at the first insertion (98 per cent vs. 86.8 per cent) and was associated with smaller samples (< 10 mg) in fewer cases (3.2 per cent vs. 4.9 per cent). Cytogenetic analysis was highly successful (99.4 per cent) and accurate; however, in one case a de novo structural rearrangement of chromosome I was not recognized. Mosaicism or rare trisomies were reported in 1.30 per cent of cases. Five diagnostic errors in DNA investigation (0.51 per cent) ended with the birth of affected fetuses. Fetal loss through 28 weeks' gestation in the pregnancies intended to continue was 2.58 per cent; the rate increased with maternal age (1.22 per cent at less than 30 years to 3.8 per cent at 40 years or more), while gestational age affected the abortion rate only at 8 weeks (odds ratio=2.22, P<0.05). Rates of premature delivery, low birth weight, and perinatal mortality did not differ from the Italian standards. By comparison with the Italian Birth Defects Registry data, no differences were found for the major malformations, including transverse limb reduction defects (TLRDs) (4.34 vs. 3.28 x 10,000). Total malformations and TLRDs did not show any pattern relation to either maternal age or gestational age.