OBJECTIVES: To compare morbidity and mortality between pediatric victims of motor vehicle collisions (MVC) who were unrestrained to those restrained and to describe compliance with child restraint usage in our population. MATERIALS AND METHODS: A retrospective consecutive chart review study was performed on MVC victims 14 years old and younger who presented to our academic, level 1 trauma emergency department in 2003. Each patient's emergency department and hospital course was reviewed and data were collected. Odds ratios (ORs) were calculated for unrestrained children with respect to restrained children for fractures; intraabdominal injuries, intrathoracic injuries, intracranial injuries, admission, surgery, blood transfusion, intubation; and deaths. Hospital charges and length of hospital stay were compared between those unrestrained and restrained. Percentage of children unrestrained was determined. RESULTS: Of 336 patients, 81 (24%) were unrestrained. Mean hospital stay for unrestrained children was longer, 1.94 days (95% confidence interval [CI] 0.75-3.12) versus 0.098 days (95% CI 0.02-0.21). Unrestrained victims had higher mean charges, $14,754 (95% CI $7676-$21,831) versus $1996 (95% CI $1207-$2786). Admissions (OR = 14.48, 95% CI 5.91-38.63), fractures (OR = 5.85, 95% CI 2.13-16.89), intraabdominal injuries (OR = 20.16, 95% CI 2.36-930.68), and intrathoracic injuries (OR = 13.09, 95% CI 1.26-647.05) were all more likely in unrestrained patients. No restrained child had intracranial injury, whereas 9/81 (11.11%) of unrestrained did. Odds were higher in unrestrained for surgery [OR = 13.09, 95% CI 3.30-74.33] and transfusion [OR = 27.61, 95% CI 3.56-229.85]. Ten out of 81 (12.35%) of unrestrained children required intubation versus none for restrained. The only 2 mortalities were unrestrained patients. CONCLUSION: Critical injuries and cost of care are higher in unrestrained than restrained children. Improved compliance with child safety restraint in southern Arizona should decrease childhood morbidity and mortality from MVCs.