Severe methylenetetrahydrofolate reductase (MTHFR) deficiency is characterized by varying degrees of developmental delay, motor and gait abnormalities, seizures, and thrombosis. Biochemical abnormalities include homocystinuria and hyperhomocysteinemia. Clinical severity correlates with MTHFR activity in cultured fibroblasts; activity can also be assayed in cultured amniocytes and chorionic villus cells (CVC). Forty-four private mutations have been identified, limiting the use of direct mutation detection for prenatal diagnosis. However, intragenic polymorphisms have been identified, making prenatal diagnosis by linkage analysis a possible option, even without knowledge of deleterious mutations. Prenatal diagnosis for severe MTHFR deficiency has been available by biochemical methodologies, but molecular genetic approaches have not yet been reported. We performed prenatal diagnosis for severe MTHFR deficiency in 11 at-risk pregnancies in seven families. A combined approach of linkage analysis and enzymatic assays was used in six pregnancies; linkage analysis alone was performed in one pregnancy. Linkage analysis for the 677C > T or 1298A > C polymorphisms predicted that all seven fetuses were unaffected. For six of these seven fetuses, enzymatic activities were also measured and demonstrated concordant results. Of the 10 pregnancies in which enzymatic assays were performed, activities in cultured amniocytes predicted six unaffected fetuses (1.4-7.1 nmol CHO/mg prot/h (U)) and one affected fetus (0.24 U [control 3.1-9.6 U]). Three pregnancies assessed via CVCs demonstrated two unaffected fetuses (3.6 and 7.7 U) and 1 affected fetus (0 U [control 4.5-7.8 U]). These values were compared to those of the probands (range = 0.02-0.7 U (control 2.4-11.7 U)) in cultured fibroblasts. Our findings suggest that linkage analysis for severe MTHFR deficiency can be a practical approach for prenatal diagnosis.