We used GPS data from paratransit vehicles to evaluate the impact of emergency vehicles on urban traffic speeds. The results indicate that speed variance is significantly higher during emergency preemption and the mean speeds of traffic flowing in the same direction as the emergency vehicle and on crossing streets are lower during preemption than during normal conditions. Regression results indicate that traffic on major arterials and traffic in the opposite direction of the emergency vehicle tend to have higher speed during signal preemption. Signal preemption during peak periods and duration of preemption had a significant negative impact on traffic speeds. Also, the transition time has a negative impact on traffic speeds. The authors recommend further research on how to optimize (minimize) the preemption duration as well as transition time. Also, the impact of median type and number of lanes should be evaluated.