First-trimester prenatal diagnosis was offered to 104 twin pregnancies mainly for advanced maternal age and cytogenetic evaluation of a new fertilization technique. Chorionic villus sampling (CVS) was performed transcervically (35%), transabdominally (23%), or by combination of these two techniques (42%). Although no placental biopsy failures occurred, two errors in fetal sexing were recorded due to non-selective placental sampling. In these two cases, both fetuses were sampled transcervically. Cytogenetic results were available for all fetuses; six of them showed an abnormal direct chromosomal pattern, but long-term villi culture analysis or additional amniocentesis (n = 1) reduced the number to four. Early fetal loss (3.4%) and perinatal mortality (6.3%) after CVS were comparable with a control group of 101 consecutive twin pregnancies without prenatal diagnosis (respectively 6.9% and 5.3%). Perinatal loss in the CVS group was associated in 10 of 12 fetuses with preterm premature rupture of the membranes and consequent preterm delivery. Mean gestational age at delivery, mean birthweight and the frequency of preterm delivery, and low birthweight infants were nearly identical in both groups. This study shows that CVS in the first trimester of pregnancy is an accurate and fast approach for prenatal diagnosis in twin gestations with an acceptable risk of adverse pregnancy outcome. However, a transcervical approach for both fetuses is not recommended.