Background Anthropometric data are important indicators of child health. This study sought to determine whether anthropometric data of extreme weight were significant predictors of perioperative morbidity in pediatric surgery. Study Design This was a cohort study of children 29 days up to 18 years of age undergoing surgical procedures at participating American College of Surgeons' NSQIP Pediatric hospitals in 2011 and 2012. The primary outcomes were composite morbidity and surgical site infection. The primary predictor of interest was weight percentile, which was divided into the following categories: ≤5th percentile, 6th to 94th, or ≥95th percentile. A hierarchical multivariate logistic model, adjusting for procedure case mix, demographic, and clinical patient characteristic variables, was used to quantify the relationship between weight percentile category and outcomes. Results Children in the ≤;5th weight percentile had 1.19-fold higher odds of overall postoperative morbidity developing than children in the nonextreme range (95% CI, 1.10-1.30) when controlling for clinical variables. Yet these children did not have higher odds of surgical site infection developing. Children in the ≥95th weight percentile did not have a significant increase in overall postoperative morbidity. However, they were at 1.35-fold increased odds of surgical site infection compared with those in the nonextreme range when controlling for clinical variables (95% CI, 1.16-1.57). Conclusions Both extremely high and extremely low weight percentile scores can be associated with increased postoperative complications after controlling for clinical variables.