Background: Some scholars posit that attempts to avert stillbirth among extremely preterm gestations may result in a live birth but an early neonatal death. The literature, however, reports no empirical test of this potential form of left truncation. We examine whether annual cohorts delivered at extremely preterm gestational ages show an inverse correlation between their incidence of stillbirth and early neonatal death. Methods: We retrieved live birth and infant death information from the California Linked Birth and Infant Death Cohort Files for years 1989 to 2015. We defined the extremely preterm period as delivery from 22 to < 28 weeks of gestation and early neonatal death as infant death at less than 7 days of life. We calculated proportions of stillbirth and early neonatal death separately by cohort year, race/ethnicity, and sex. Our correlational analysis controlled for well-documented declines in neonatal mortality over time. Results: California reported 89,276 extremely preterm deliveries (live births and stillbirths) to Hispanic, non-Hispanic (NH) Black, and NH white mothers from 1989 to 2015. Findings indicate an inverse correlation between stillbirth and early neonatal death in the same cohort year (coefficient:-0.27, 95% CI of − 0.11; − 0.42). Results remain robust to alternative specifications and falsification tests. Conclusions: Findings support the notion that cohorts with an elevated risk of stillbirth also show a reduced risk of early neonatal death among extremely preterm deliveries. Results add to the evidence base that selection in utero may influence the survival characteristics of live-born cohorts.