Abstract Background The quantification of alpha-fetoprotein (AFP) in maternal serum is commonly used in prenatal screening to estimate the risk of fetal open neural tube defects (ONTDs). This involves comparing a patient’s measured AFP concentration to the median AFP for women of similar gestational age and weight, reported in AFP multiples of the median (MoM). Current guidelines recommend establishing population medians separately grouped by patients’ race, based on historical reports documenting differences in median AFP concentrations across different populations. We aimed to reevaluate this practice of race based ONTD risk stratification given the limited and questionable historical evidence as well as the small observed differences in AFP concentrations in our practice and in a recent report. Methods We collected maternal serum AFP screening data for patients between 12/30/19 and 2/9/24. Informed by the statistical methods used for clinical risk stratification, we developed two log-linear multiple variable regression models for estimating AFP: gestational age (GA), weight, and interaction terms (Model 1) and Model 1 predictors, race, and interaction terms (Model 2). AFP MoMs were calculated by dividing the measured AFP concentration by the model-estimated AFP. Results We analyzed maternal ONTD screening for 7,702 patients with singleton pregnancies. The current clinical process uses race and ethnicity as documented in the electronic health record, which was categorized as Black (N = 2,360), White (N = 3,734), Hispanic (N = 209), Asian (N = 641), and Other (N = 758). The addition of race in Model 2 included statistically significant predictors (p < 0.001) for Black (β = 0.2), Asian (β = 0.05), and Black:weight interaction (β = -5.9 x 10-4). Between Model 1 and Model 2, the mean change in AFP MoMs was -0.05, +0.03, 0.0, -0.02, and +0.01 for Black, White, Hispanic, Asian, and Other patients, respectively. 95% of Black patients had a change < 0.12 in AFP MoM and > 99% of all other patients had a change < 0.1. Using the clinical interpretative threshold, 56 (0.7%) results were considered elevated in both models, 4 (0.05%) elevated in Model 1 only (3 Black, 1 Other), and 4 (0.05%) elevated in Model 2 only (3 White, 1 Hispanic). Chart review of the 8 discordant interpretations showed small changes in MoMs (mean absolute difference = 0.09) and no cases of ONTDs. Conclusions In our health system’s population, maternal race appears to have a small statistically significant association with maternal AFP concentration, even after adjustment for weight and maternal GA. However, preliminary analyses indicate that this difference has a small impact on the clinical interpretation of these studies, with most patients classified similarly when race is not accounted for in the calculation. Further investigation is needed to clarify whether this small impact on estimation of risk warrants the continued inclusion of race in NTD screening, as well as whether other clinical, social, or community factors available for inclusion in this clinical risk stratification model would be better predictors than the poorly defined and discriminatory concept of race.