SUMMARYWe measured Concanavalin A (Con A) non-binding alphafetoprotein in amniotic fluid from 21 normal pregnancies and 20 abnormal pregnancies (complicated by anencephaly, spina bifida, or exomphalos) using small Con A-sepharose chromatography columns. There was a highly significant difference between percentage non-binding alphafetoprotein levels for pregnancies with a normal outcome (mean result 32 %; range 18--47 %) and pregnancies complicated by a neural tube defect or exomphalos (mean 14%; range 7-20%). The test is therefore of potential value in cases where there is uncertainty over the interpretation of total amniotic fluid alphafetoprotein levels.The measurement of alphafetoprotein (AFP) in amniotic fluid is now widely used as a follow-up to serum AFP screening for the detection of neural tube defects such as spina bifida and anencephaly. When a cut-off level of five standard deviations above the mean for normal pregnancies is taken, amniotic fluid AFP estimations give a false-positive rate of 0·2% but may miss 10% of open spina bifida cases.' Thus there is a need for additional diagnostic procedures which would aid the interpretation of 'borderline' amniotic fluid AFP results, for example, those cases that fall between 2· 5 and 5 standard deviations above the mean.AFP, a glycoprotein containing approximately 4 % carbohydrate, is composed of two molecular variants which differ only in their carbohydrate content.s Affinity chromatography using Concanavalin A (Con A) will separate amniotic fluid AFP into two components: Can A non-binding AFP, which passes through the chromatography column, and a Con A binding fraction, which is retained.The proportion of Con A non-binding AFP in fetal serum and amniotic fluid: 2-6 % of fetal serum AFP and 14-45% of amniotic fluid AFP is not bound by Con A. 3