We sought to determine the reliability of the Glasgow Coma Scale (GCS) when used by emergency physicians and paramedics. We performed a prospective sequential trial in a classroom setting, with subjects blinded to others' scoring. Nineteen university-affiliated emergency physicians and 41 professional paramedics from an urban EMS system voluntarily participated. Participants viewed four videotaped scenes in which a patient is assessed by a paramedic. The first three scenes represented severe, intermediate, and no/mild alteration in level of consciousness (LOC). The findings in the fourth scene were identical to the first, allowing determination of intrarater reliability. The Kappa statistic was used to determine interrater reliability; the reliability coefficient determined intrarater reliability. Kappa was significant (p < 0.0001) for severe (k = 0.48), intermediate (K = 0.34), and no/mild (k = 0.85) conditions. Intrarater reliability (r1,2) for emergency physicians was 0.66 (p < 0.01) and for paramedics was 0.63 (p < 0.01). The GCS shows statistically significant reliability (i.e., significant agreement) between emergency physicians and emergency medical technician-paramedics. It also has a significant level of interpreter reliability.