Background: Radiation dose metrics vary by the calibration reference phantom used to report doses. By convention, 16-cm diameter cylindrical polymethyl-methacyrlate phantoms are used for head imaging and 32-cm diameter phantoms are used for body imaging in adults. Actual usage patterns in children remain under-documented. Objective: This study uses the University of California San Francisco International CT Dose Registry to describe phantom selection in children by patient age, body region and scanner manufacturer, and the consequent impact on radiation doses. Materials and methods: For 106,837 pediatric computed tomography (CT) exams collected between Jan. 1, 2015, and Nov. 2, 2020, in children up to 17 years of age from 118 hospitals and imaging facilities, we describe reference phantom use patterns by body region, age and manufacturer, and median and 75th-percentile dose–length product (DLP) and volume CT dose index (CTDIvol) doses when using 16-cm vs. 32-cm phantoms. Results: There was relatively consistent phantom selection by body region. Overall, 98.0% of brain and skull examinations referenced 16-cm phantoms, and 95.7% of chest, 94.4% of abdomen and 100% of cervical-spine examinations referenced 32-cm phantoms. Only GE deviated from this practice, reporting chest and abdomen scans using 16-cm phantoms with some frequency in children up to 10 years of age. DLP and CTDIvol values from 16-cm phantom-referenced scans were 2–3 times higher than 32-cm phantom-referenced scans.