BACKGROUND: Estimates of blood loss during craniosynostosis surgeries have ranged from 42 to 126 percent of infant's total blood volume. Currently, no risk model has been developed to determine the likelihood of needing a blood transfusion. The authors propose an objective model, based on patients' three-dimensional anatomical characteristics, to stratify the likelihood of needing perioperative packed red blood cells. METHODS: High-throughput image analysis from already ordered preoperative computed tomographic scans was used to reconstruct the three-dimensional anatomy of the temporalis muscle and overlying temporal fat pad. Using these morphomic measurements, the authors created a risk assessment model of the amount of packed red blood cells infused based on morphomic variations in temporal soft tissue. RESULTS: The authors evaluated 139 infants with nonsyndromic craniosynostosis from the University of Michigan Health System. Results show the need for blood transfusion ranged from 94.1 percent among patients in the smallest quartile for temporal fat pad volume compared with 65.7 percent among the group with the largest quartile for temporal fat pad volume (p = 0.0057). Using multivariable linear regression, temporal fat pad volume (p = 0.012) and fat pad thickness (p = 0.036) were independent predictors for amount of packed red blood cells required. CONCLUSIONS: The authors demonstrate that patients with diminished temporal fat pad volume are significantly more likely to need increased packed red blood cell transfusions. In addition, by use of multivariable linear regression, their data suggest that temporal fat pad volume and thickness were independent predictors for the amount of required transfusion of packed red blood cells.