During pregnancy, fetal cells suitable for genetic testing can be obtained from amniotic fluid by amniocentesis (AC), placental tissue by chorionic villus sampling (CVS), or fetal blood. A major disadvantage of second trimester amniocentesis is that the results are available relatively late in pregnancy (after 16 weeks' gestation). Earlier alternatives are chorionic villus sampling (CVS) and early amniocentesis, which can be performed in the first trimester of pregnancy. The objective of this review was to compare the safety and accuracy of all types of AC (i.e. early and late) and CVS (e.g. transabdominal, transcervical) for prenatal diagnosis. We searched the Cochrane Pregnancy and Childbirth Group's Trials Register (3 March 2017), ClinicalTrials.gov, the WHO International Clinical Trials Registry Platform (ICTRP; 3 March 2017), and reference lists of retrieved studies. All randomised trials comparing AC and CVS by either transabdominal or transcervical route. Two review authors independently assessed trials for inclusion and risk of bias, extracted data and checked them for accuracy. The quality of the evidence was assessed using the GRADE approach. We included a total of 16 randomised studies, with a total of 33,555 women, 14 of which were deemed to be at low risk of bias. The number of women included in the trials ranged from 223 to 4606.Studies were categorized into six comparisons: 1. second trimester AC versus control; 2. early versus second trimester AC; 3. CVS versus second trimester AC; 4. CVS methods; 5. Early AC versus CVS; and 6. AC with or without ultrasound.One study compared second trimester AC with no AC (control) in a low risk population (women = 4606). Background pregnancy loss was around 2%. Second trimester AC compared to no testing increased total pregnancy loss by another 1%. The confidence intervals (CI) around this excess risk were relatively large (3.2% versus 2.3 %, average risk ratio (RR) 1.41, 95% CI 0.99 to 2.00; moderate-quality evidence). In the same study, spontaneous miscarriages were also higher (2.1% versus 1.3%; average RR 1.60, 95% CI 1.02 to 2.52; high-quality evidence). The number of congenital anomalies was similar in both groups (2.0% versus 2.2%, average RR 0.93, 95% CI 0.62 to 1.39; moderate-quality evidence).One study (women = 4334) found that early amniocentesis was not a safe early alternative compared to second trimester amniocentesis because of increased total pregnancy losses (7.6% versus 5.9%; average RR 1.29, 95% CI 1.03 to 1.61; high-quality evidence), spontaneous miscarriages (3.6% versus 2.5%, average RR 1.41, 95% CI 1.00 to 1.98; moderate-quality evidence), and a higher incidence of congential anomalies, including talipes (4.7% versus 2.7%; average RR 1.73, 95% CI 1.26 to 2.38; high-quality evidence).When pregnancy loss after CVS was compared with second trimester AC, there was a clinically significant heterogeneity in the size and direction of the effect depending on the technique used (transabdominal or transcervical), therefore, the results were not pooled. Only one study compared transabdominal CVS with second trimester AC (women = 2234). They found no clear difference between the two procedures in the total pregnancy loss (6.3% versus 7%; average RR 0.90, 95% CI 0.66 to 1.23, low-quality evidence), spontaneous miscarriages (3.0% versus 3.9%; average RR 0.77, 95% CI 0.49 to 1.21; low-quality evidence), and perinatal deaths (0.7% versus 0.6%; average RR 1.18, 95% CI 0.40 to 3.51; low-quality evidence). Transcervical CVS may carry a higher risk of pregnancy loss (14.5% versus 11.5%; average RR 1.40, 95% CI 1.09 to 1.81), but the results were quite heterogeneous.Five studies compared transabdominal and transcervical CVS (women = 7978). There were no clear differences between the two methods in pregnancy losses (average RR 1.16, 95% CI 0.81 to 1.65; very low-quality evidence), spontaneous miscarriages (average RR 1.68, 95% CI 0.79 to 3.58; very low-quality evidence), or anomalies (average RR 0.68, 95% CI 0.41 to 1.12; low-quality evidence). We downgraded the quality of the evidence to low due to heterogeneity between studies. Transcervical CVS may be more technically demanding than transabdominal CVS, with more failures to obtain sample (2.0% versus 1.1%; average RR 1.79, 95% CI 1.13 to 2.82, moderate-quality evidence).Overall, we found low-quality evidence for outcomes when early amniocentesis was compared to transabdominal CVS. Spontaneous miscarriage was the only outcome supported by moderate-quality evidence, resulting in more miscarriages after early AC compared with transabdominal CVS (2.3% versus 1.3%; average RR 1.73, 95% CI 1.15 to 2.60). There were no clear differences in pregnancy losses (average RR 1.15, 95% CI 0.86 to 1.54; low-quality evidence), or anomalies (average RR 1.14, 95% CI 0.57 to 2.30; very low-quality evidence).We found one study that examined AC with or without ultrasound, which evaluated a type of ultrasound-assisted procedure that is now considered obsolete. Second trimester amniocentesis increased the risk of pregnancy loss, but it was not possible to quantify this increase precisely from only one study, carried out more than 30 years ago.Early amniocentesis was not as safe as second trimester amniocentesis, illustrated by increased pregnancy loss and congenital anomalies (talipes). Transcervical chorionic villus sampling compared with second trimester amniocentesis may be associated with a higher risk of pregnancy loss, but results were quite heterogeneous.Diagnostic accuracy of different methods could not be assessed adequately because of incomplete karyotype data in most studies.