Concanavalin A (con A) and lens culinaris agglutinin (LCA) microheterogeneity pattern of AFP (crossed affinity immunoelectrophoresis), alpha-2-macroglobulin and synaptic membrane protein D-2 (rocket immunoelectrophoresis) and qualitative (polyacrylamide gel electrophoresis) and quantitative (enzyme kinetic reaction rate) acetylcholinesterase were analysed in 87 consequtive samples from normal pregnancies and 37 abnormal samples (fetal neural tube defect or abdominal wall defect). Very few false positive results were obtained in normal pregnancies with any of the tests. In all cases of neural tube defects the correct result was obtained with qualitative acetylcholinesterase analysis, whereas only 2/3 of the abdominal wall defects were correctly predicted. Testing with con A or LCA was less optimal in neural tube defects, whereas all abdominal wall defects could be predicted correctly. Acetylcholinesterase in the quantitative test and protein D-2 did not decrease the rate of false results. Determination of the alpha-2-macroglobulin concentration performed well in the present study, but is not recommended because of the very high susceptibility to contamination of amniotic fluid with fetal or maternal blood.