Cytogenetic analysis was performed in 4860 chorionic villus samples by means of both direct preparation and long-term culture. The results of the analysis were compared with a classification including all theoretical types of combinations between the chromosomal constitution of the cytotrophoblast, extraembryonal mesoderm, and fetus, with the aim of evaluating the cytogenetic variability along the trophoblast-embryo axis. Eighteen of 29 possible combinations were found demonstrating a considerable heterogeneity. A mosaic conceptus was found in 1.5 per cent of cases, with generalized mosaicisms and confined mosaicisms in 0.2 and 1.3 per cent, respectively. Cytogenetic variability along the trophoblast-embryo axis was found in 1.42 per cent of cases. Results possibly leading to diagnostic errors (false-positive and false-negative results) were found in only 1.38 per cent. False-positive results of direct preparation were the most commonly observed discrepancy (0.8 per cent), while the incidence of false-positive results of the culture method and of both methods was 0.31 and 0.16 per cent respectively. The incidence of false-negative results was 0.1 per cent, with false-negative results of direct preparation 0.08 per cent and false-negative results of both methods 0.02 per cent. False-negative results of the culture method were not found. Our data confirm the high diagnostic accuracy of chorionic villus sampling and the utility of the combined use of the two methods in minimizing diagnostic errors and in reducing the need for follow-up amniocentesis.