A (semi-) randomized controlled study with long-term follow-up was conducted to compare the effects of transabdominal chorionic villus sampling and early amniocentesis on fetal mortality and child morbidity. Women requesting early prenatal diagnosis for advanced maternal age were allocated to early amniocentesis or transabdominal chorionic villus sampling either by randomization or, if they declined randomization, by their own choice. Of the 212 women who entered the study, 117 were randomized, 70 chose early amniocentesis and 25 chose transabdominal chorionic villus sampling. Overall, 130 women underwent early amniocentesis and 74 underwent transabdominal chorionic villus sampling at a median gestation of 12 weeks. Two women were excluded because of fetal death before the procedure. Mosaic karyotypes were found in 5.4 per cent of the early amniocenteses and in none of the chorionic villus samples. All unintended fetal losses occurred after early amniocentesis with a frequency of 6.2 per cent (95 per cent confidence interval: 2.7 per cent to 11.8 per cent). Talipes equinovarus was only observed after early amniocentesis with a frequency of 3.1 per cent (95 per cent confidence interval: 0.8 per cent to 7.7 per cent). We conclude that chorionic villus sampling remains the method of choice if prenatal diagnosis is needed in the first trimester of pregnancy.