A hypothesized inverse relationship between the concentration of glycosylated hemoglobin and serum alpha-fetoprotein was observed in a prospective study of 39 pregnant women with insulin-dependent diabetes as well as seven pregnant women with diabetes who did not require insulin (r = -0.434, p less than 0.002). No similar correlation was found among a selected population of healthy pregnant women (r = -0.129). Because the level of glycosylated hemoglobin in pregnancy correlates with poor outcome, including the occurrence of fetal anomaly, it may be important to quantify glycosylated hemoglobin in pregnancies with low alpha-fetoprotein levels. These results also suggest that the maternal concentration of glycosylated hemoglobin can be used to adjust serum alpha-fetoprotein values before their interpretation in the screening of pregnant women with diabetes.