A retrospective study was performed to compare the detection rate of chromosomal abnormalities by different approaches of full karyotyping, rapid aneuploidy diagnosis (RAD) or both when invasive prenatal testing is performed for diagnosis of thalassaemia. The karyotype results of 1120 prenatal samples obtained from thalassaemia couples from January 1985 to December 2002 in a referral centre for prenatal diagnosis were studied. The detection rate of chromosomal abnormalities by four different approaches were compared: (i) karyotyping for all samples; (ii) RAD (21,18,13,X,Y) for all samples; (iii) RAD for all samples + karyotyping for cases with ultrasound abnormalities; and (iv) RAD (21,18,13) for all + RAD (X,Y) for cases with ultrasound abnormalities consistent with Turner syndrome + karyotyping for cases with ultrasound abnormalities. Normal karyotypes were found in 1103 samples (98.5%). There were 17 cases (1.5%) of chromosomal abnormalities: four cases (0.36%) were clinically significant, eight cases (0.7%) were of borderline clinical significance and five cases (0.44%) were not confirmed by subsequent prenatal or postnatal tests. The incidences of autosomal (7/1120 = 0.63%) and sex chromosomal (5/1120 = 0.45%) abnormalities were not higher than those (0.41 and 0.22%, respectively) from newborn surveys (Hook and Hamerton, 1977) (P = 0.398 and 0.216, respectively). Approach 1 would detect all 17 chromosomal abnormalities. Approach 2 would detect three of four clinically significant chromosomal abnormalities but not detect six of eight chromosomal abnormalities of borderline clinical significance and three of five chromosomal abnormalities not confirmed by subsequent prenatal or postnatal tests. Approach 3, in addition, would be able to detect all four clinically significant chromosomal abnormalities. Approach 4 would detect all four clinically significant chromosomal abnormalities but would not detect seven of eight chromosomal abnormalities of borderline clinical significance and four of five chromosomal abnormalities not confirmed by subsequent prenatal or postnatal tests. RAD (21,18,13) for all + RAD (X,Y) for cases with ultrasound abnormalities consistent with Turner syndrome + karyotyping for cases with ultrasound abnormalities seemed to be the best approach for the detection of chromosomal abnormalities when invasive prenatal testing is performed for diagnosis of thalassaemia.