Abstract. Emergency medical services (EMS) systems increasingly seek to triage patients to alternative EMS resources. Emergency medical services dispatchers may be asked to perform this triage. New protocols may be necessary. Alternatively, existing protocols may be sufficient for this task. For an existing dispatch protocol to be sufficient, it at least must accurately categorize patient condition and severity based on an external standard. Objective: To examine the extent to which nature codes (NCs), or patient condition codes, and severity codes (SCs) currently assigned in one urban 911 center agree with paramedic field findings. The null hypothesis was that there is no routine agreement (75%) between dispatcher-assigned NC or SC and paramedic-assigned NC or SC for the same patient using the same protocol. Methods: Emergency medical services dispatch nature and severity code data and matching out-ofhospital data were prospectively gathered over six months. Dispatch data included the NC: caller-identified problem, and the SC: dispatcher-assessed severity. Each NC is modified by one of three SCs (1, 3, or 9): 1 is emergent, 3 is urgent, and 9 is neither. Paramedics verified and/or corrected dispatcher-assigned NCs and SCs using the same dispatch protocol. Results: One thousand forty usable cases fell into 33 unique NC/SC combinations. The designation of SC 1 was assigned 275 times, SC 3 was assigned 736 times, and SC 9 was assigned 24 times. The SC was missing five times. The overall NC agreement was 0.70 (95% CI = 0.697 to 0.703). The overall SC agreement was 0.65 (95% CI = 0.645 to 0.655). The NC agreement exceeded 75% for ten (59%) NC/SC combinations. The SC agreement exceeded 75% for five (29%) NC/SC combinations. There was both NC and SC agreement for four (24%) combinations: urgent breathing problems, urgent diabetic problems, urgent falls, and urgent overdoses. The greatest NC/SC disagreement occurred within emergent and urgent traffic crashes. Paramedics adjusted SC toward lower severity 29% of the time and toward higher severity 5.4% of the time. There was no upward SC adjustment for eight (47%) combinations. Conclusions: Certain dispatcher-assigned NC and SC codes and NC/ SC combinations achieved the study threshold. Overall agreement failed to achieve the threshold. The lowest SC level was rarely assigned, preventing a meaningful analysis of all severity levels.