We investigate the impact of early-life medical interventions on low-risk newborn health. A policy rule in The Netherlands creates large discontinuities in medical treatments at gestational week 37. Using a regression discontinuity design, we find no health benefits from additional treatments for average newborns. However, there is substantial heterogeneity in returns to treatments with significant health benefits for newborns in the lowest income quartile and no benefits in higher income quartiles. This seems due to increased maternal stress from referral to an obstetrician among higher-income mothers, heterogeneous effects of home births, and potential difficulties in risk screening among low-income women.