Prenatal diagnosis of cystic fibrosis by microvillar enzyme assay on amniotic fluid supernatant has been carried out on 258 sequential pregnancies with a 1 in 4 recurrence risk, all with known outcome. In general the three enzymes evaluated, gamma-glutamyltranspeptidase, aminopeptidase M and the intestinal isoenzyme of alkaline phosphatase, showed a high degree of concordance. However, there were two unusual patterns of microvillar enzyme activity; in seven cases a low gamma-glutamyltranspeptidase activity was associated with elevated values of intestinal alkaline phosphatase, and in ten cases there were isolated low values of intestinal alkaline phosphatase. The former pattern was found to be associated with cystic fibrosis in five cases, while the latter was associated with a normal outcome in all ten cases. A retrospective analysis of enzyme values suggested that the optimal system for minimizing false positives and false negatives was to define foetal cystic fibrosis as a sample where two of the three microvillar enzymes were below a cut-off of half the median value for the gestational week. If such scoring were applied to the cases where conventional microvillar enzyme patterns were observed, the false positive rate was 2.3% and the false negative rate 4.4% between 17 and 20 weeks of gestation.