Heart rate variability (HRV) is an important biomarker for parasympathetic function and future health outcomes. The present study examined how the structure of regions in a neural network thought to maintain top-down control of parasympathetic function is associated with HRV during both rest and social stress. Participants were 127 young women (90 Black American), who completed a structural MRI scan and the Trier Social Stress Test (TSST), during which heart rate was recorded. Regression analyses were used to evaluate associations between cortical thickness in five regions of the Central Autonomic Network (CAN; anterior midcingulate cortex [aMCC], pregenual and subgenual anterior cingulate cortex [pgACC, sgACC], orbitofrontal cortex [OFC], and anterior insula) and high-frequency HRV during rest and stress. Results indicated that cortical thickness in CAN regions did not predict average HRV during rest or stress. Greater cortical thickness in the right pgACC was associated with greater peak HRV reactivity during the TSST, and survived correction for multiple comparisons, but not sensitivity analyses with outliers removed. The positive association between cortical thickness in the pgACC and peak HRV reactivity is consistent with the direction of previous findings from studies that examined tonic HRV in adolescents, but inconsistent with findings in adults, which suggests a possible neurodevelopmental shift in the relation between brain structure and autonomic function with age. Future research on age-related changes in brain structure and autonomic function would allow a more thorough understanding of how brain structure may contribute to parasympathetic function across neurodevelopment.