Prenatal diagnosis has been undertaken in 17 pregnancies in 15 families at risk for aspartoacylase deficiency. Amniocentesis was at 14-18 weeks gestation followed by measurement of amniotic fluid N-acetyl-L-aspartate (NAA) levels in all pregnancies and amniocyte aspartoacylase activity in most pregnancies. In one case amniocentesis was performed at 11 weeks gestation in conjunction with chorionic villus sampling. At 14-18 weeks of gestation, control levels of NAA were 0.30-2.55 mumol/L. The fetus was predicted to be affected in 8 of the pregnancies, 4 of which were confirmed by enzyme analysis on fetal tissue and 2 by the clinical and metabolic expression of Canavan disease in a newborn. In two cases there was no fetal tissue available for enzyme confirmation. One of these had the highest amniotic fluid NAA level (8.68 mumol/L) and in the other pregnancy there were two amniocenteses, both with markedly elevated levels. Of 9 fetuses predicted to be normal, 8 newborns were clinically and biochemically normal. A single case with amniotic fluid NAA in the normal range (1.56 mumol/L, measured in one laboratory only) resulted in an aborted fetus in whom aspartoacylase was deficient in cultured skin fibroblasts. We propose that amniotic fluid NAA levels remain the best predictor of an affected fetus and recommend that the assay be performed in multiple laboratories.