Prenatal caffeine exposure (PCE) has been positively associated with elevated body mass index (BMI) in children. Why this association occurs is unclear, but it is possible that PCE alters the in utero development of brain structures associated with food preference, leading to more total sugar intake (TSI, grams) later in childhood. To test this hypothesis, we investigated if PCE (daily/weekly/<weekly vs. no exposure) and elevated BMI are associated with increased TSI, neural activation during large reward anticipation (monetary incentive delay task—functional MRI) and structural changes (thickness, mm) in taste processing regions of children (n = 5534; 9−11 years) from the large-scale Adolescent Brain Cognitive Development (ABCD) study. Linear mixed-effect models, after covariate adjustments, identified a positive association (p < 0.05, all |βs| > 0.01) of excessive PCE (vs. no exposure) with elevated BMI (daily/weekly/daily limit; consistent in boys and girls), increased TSI (daily) and insular thickness (daily/weekly), as well as low middle frontal cortex (MFC) activation (daily). Our sub-analysis revealed an association of daily/weekly PCE (vs. no exposure) with increased gram sugar intake from soft drinks. We also identified a positive relationship of excessive PCE with elevated TSI and increased insular thickness (a key gustatory region), while in a Sobel test, reward sensitivity (reduced brain reactivity to reward anticipation in MFC; tracks reward outcomes) mediated (Test statistic = 2.23; p = 0.02) the PCE-linked BMI changes in adolescents. Our findings suggest that excessive PCE might be detrimental to frontal lobe development and altered reward sensitivity to food, thereby increasing risk for elevated TSI and obesity. Our results support recommendations to limit caffeine intake during pregnancy.