Amniotic fluid (AF) levels of 17-hydroxyprogesterone (17OHP) and testosterone (T) were determined at 16-17 weeks in 17 pregnancies at risk for CAH and results compared to 75 normal controls. The fetus was predicted to be unaffected in 12 cases on the findings of normal AF levels of both 17OHP and T and the latter allowed a correct prediction of fetal sex in all instances. HLA typing confirmed normality in 12 cases revealing 5 carriers, 5 homozygous normal and 2 indeterminate. Steroid levels of the 2 groups were similar. Three fetuses were predicted to be CAH affected on unambiguously high levels of 17OHP and T (in female only). HLA typing was in agreement, and the diagnosis was confirmed in 2 abortuses and a female newborn by physical and hormonal studies. In the last 2 cases AF levels of OHP and T were normal but HLA (A/B/C) genotypes were identical to the CAH affected siblings. Normal physical and hormonal findings in the 2 aborted fetuses would exclude the possibility of an in utero virilizing form of CAH. The discrepancy could be explained on the basis that the fetuses had an allelic form of 21-hydroxylase deficiency or on the basis of recombination (not fully tested). It is concluded that a fully informative prenatal diagnosis of CAH should not rely entirely on HLA typing but on hormonal studies.