We report cytogenetic results from a randomized Danish chorionic villus sampling (CVS) and amniocentesis (AC) study including 2928 placental and 1075 amniotic fluid specimens processed in the same laboratory. The results are presented in groups comparing CVS with amniocentesis and transabdominal (TA) CVS with transcervical (TC) CVS as randomized. More abnormalities and more ambiguous diagnostic problems were found in placental tissues than in amniotic cells. There were no diagnostic errors and no incorrect sex predictions. Mosaicism was detected in 1 per cent of all cases of CVS (discordancies included). When confirmation studies were done, 90 per cent were found to be confined to the placenta. Eight cases (0.7 per cent) of mosaicism/pseudomosaicism were seen in amniotic fluid specimens, and two cases of five with confirmation studies were confirmed in the fetus. The rate of mosaicism/pseudomosaicism in CVS and AC specimens differed (p < 0.05). The rate of pseudomosaicism in cultures of villi and amniotic fluid cells was 0.5 and 0.6 per cent, respectively. Single-cell aneuploidy was observed in 1.8 per cent of villi and 1.4 per cent of amniotic fluid cell specimens. Maternal cell contamination (MCC) was seen more often after TC sampling (4.5 per cent) compared with TA sampling (1.5 per cent), but posed no problems in interpretation. Compared with the processing of cultured specimens, the short-term method of preparation of villi in our laboratory doubled the technicians' workload. For practical and economic reasons we have ceased the routine use of short-term preparations.