ImportanceInequalities in preterm infant mortality exist between population subgroups within the United States.ObjectiveTo characterize trends in preterm infant mortality by maternal race and socioeconomic status to assess how inequalities in preterm mortality rates have changed over time.Design, Setting, and ParticipantsThis was a retrospective longitudinal descriptive study using the US National Center for Health Statistics birth infant/death data set for 12 256 303 preterm infant births over 26 years, between 1995 and 2020. Data were analyzed from December 2022 to March 2023.ExposuresMaternal characteristics including race, smoking status, educational attainment, antenatal care, and insurance status were used as reported on an infant’s US birth certificate.Main Outcomes and MeasuresPreterm infant mortality rate was calculated for each year from 1995 to 2020 for all subgroups, with a trend regression coefficient calculated to describe the rate of change in preterm mortality.ResultsThe average US preterm infant mortality rate (IMR) decreased from 33.71 (95% CI, 33.71 to 34.04) per 1000 preterm births per year between 1995-1997, to 23.32 (95% CI, 23.05 to 23.58) between 2018-2020. Black non-Hispanic infants were more likely to die following preterm births than White non-Hispanic infants (IMR, 31.09; 95% CI, 30.44 to 31.74, vs 21.81; 95% CI, 21.43 to 22.18, in 2018-2020); however, once born, extremely prematurely Black and Hispanic infants had a narrow survival advantage (IMR rate ratio, 0.87; 95% CI, 0.84 to 0.91, in 2018-2020). The rate of decrease in preterm IMR was higher in Black infants (−0.015) than in White (−0.013) and Hispanic infants (−0.010); however, the relative risk of preterm IMR among Black infants compared with White infants remained the same between 1995-1997 vs 2018-2020 (relative risk, 1.40; 95% CI, 1.38 to 1.44, vs 1.43; 95% CI, 1.39 to 1.46). The rate of decrease in preterm IMR was higher in nonsmokers compared with smokers (−0.015 vs −0.010, respectively), in those with high levels of education compared with those with intermediate or low (−0.016 vs – 0.010 or −0.011, respectively), and in those who had received adequate antenatal care compared with those who did not (−0.014 vs −0.012 for intermediate and −0.013 for inadequate antenatal care). Over time, the relative risk of preterm mortality widened within each of these subgroups.Conclusions and RelevanceThis study found that between 1995 and 2020, US preterm infant mortality improved among all categories of prematurity. Inequalities in preterm infant mortality based on maternal race and ethnicity have remained constant while socioeconomic disparities have widened over time.