Background: In critically ill children, omitting early use of parenteral nutrition (late-PN versus early-PN) reduced infections, accelerated weaning from mechanical ventilation, and shortened PICU stay. We hypothesized that fasting-induced ketogenesis mediates these benefits. Methods: In a secondary analysis of the PEPaNIC RCT (N = 1440), the impact of late-PN versus early-PN on plasma 3-hydroxybutyrate (3HB), and on blood glucose, plasma insulin, and glucagon as key ketogenesis regulators, was determined for 96 matched patients staying ≥ 5 days in PICU, and the day of maximal 3HB-effect, if any, was identified. Subsequently, in the total study population, plasma 3HB and late-PN-affected ketogenesis regulators were measured on that average day of maximal 3HB effect. Multivariable Cox proportional hazard and logistic regression analyses were performed adjusting for randomization and baseline risk factors. Whether any potential mediator role for 3HB was direct or indirect was assessed by further adjusting for ketogenesis regulators. Results: In the matched cohort (n = 96), late-PN versus early-PN increased plasma 3HB throughout PICU days 1-5 (P < 0.0001), maximally on PICU day 2. Also, blood glucose (P < 0.001) and plasma insulin (P < 0.0001), but not glucagon, were affected. In the total cohort (n = 1142 with available plasma), late-PN increased plasma 3HB on PICU day 2 (day 1 for shorter stayers) from (median [IQR]) 0.04 [0.04-0.04] mmol/L to 0.75 [0.04-2.03] mmol/L (P < 0.0001). The 3HB effect of late-PN statistically explained its impact on weaning from mechanical ventilation (P = 0.0002) and on time to live PICU discharge (P = 0.004). Further adjustment for regulators of ketogenesis did not alter these findings. Conclusion: Withholding early-PN in critically ill children significantly increased plasma 3HB, a direct effect that statistically mediated an important part of its outcome benefit.