Objective To examine the differences in multiples of the median (MoM) of total alpha-fetoprotein, and the proportion of Lens culinaris agglutinin reactive alpha-fetoprotein (% alpha-fetoprotein-L2 1 L3) in the maternal serum and amniotic fluid of pregnant women whose fetuses were diagnosed with autosomal or sex chromosomal abnormalities. Design Prospective consecutive series. Setting University hospital.Sample Maternal sera and amniotic fluids from 46 pregnant women with trisomy 21 fetuses, 10 pregnant women with trisomy 18 fetuses, one pregnant woman with a trisomy 13 fetus, six pregnant women with fetal sex chromosomal abnormalities, and 100 pregnant women for whom the fetal karyotype was diagnosed as normal following a genetic amniocentesis. Results The proportion of alpha-fetoprotein-L2 1 L3 in maternal serum for trisomy 21 (40.3%, P , 0.0001) and trisomy 18 (39.8%, P , 0.05) showed a significantly higher value compared with normal (32.6%). The proportion of alpha-fetoprotein-L2 1 L3 in amniotic fluid was significantly higher (P , 0.0001) for trisomy 21 (46.6%) than for a normal karyotype (41.5%). Only for the trisomy 21 group was there a strong correlation in the % alpha-fetoprotein-L2 1 L3 between maternal serum and amniotic fluid (r ¼ 0.840, P , 0.0001). For all groups, there was no correlation between alpha-fetoprotein MoM and % alpha-fetoprotein-L2 1 L3 in maternal serum and amniotic fluid. Conclusion The proportion of alpha-fetoprotein-L2 1 L3 in maternal serum is an appropriate choice for a trisomy 21 biochemical marker, and it is possible that combining alpha-fetoprotein-L2 1 L3 analysis with assays of alpha-fetoprotein in maternal serum could further improve the sensitivity and specificity of multiple marker screening.