BACKGROUND: The Glasgow Coma Scale (GCS) score is widely accepted as the cornerstone in the assessment of neurologic function after trauma and is being used to select patients for enrollment in prehospital (Ph) intervention trials. This study was performed to assess the degree of agreement between Ph-GCS and emergency department (ED)-GCS scores in our trauma system patients. METHODS: From January 1, 2000, through November 30, 2003, 3,669 Emergency Medical System records were linked to our institution's trauma registry. Patients were classified according to their Ph-GCS and ED-GCS scores: 13 to 15, mild; 9 to 12, moderate; and 3 to 8, severe injury. Weighted kappa-coefficients and their 95% confidence intervals were used to examine the agreement between Ph-GCS and ED-GCS patient classifications. RESULTS: Percent agreement was high for the mild category (97.9%) but diminished for moderate (9.3%) and severe (63.3%) groups. Overall, the kappa-coefficient was 0.53 (95% confidence interval 0.48-0.58), indicating moderate agreement. By GCS component, kappa-coefficients were similar, ranging from 0.52 for the verbal component to 0.48 for the eye and motor components. kappa-values decreased even further for those with longer (i.e. >or=20 minutes) transport times. CONCLUSION: In this study, differences between Ph-GCS and ED-GCS were seen in the patient population with moderate to severe head injury. Individual component score agreement was moderate for all categories, with agreement worsening for patients with longer transport times. Although this suggests improvement in the patient's condition during transport, inter-rater variability between Ph and ED personnel cannot be excluded. The utility of Ph-GCS in Ph interventional trials for determining the presence of significant head injury will need to be further evaluated.