This paper presents a new method for predicting unimpeded taxi time based on the airport node-link model through statistical analysis of the airport surface detection equipment surveillance data. The proposed method can predict taxi-out and taxi-in times for departure and arrival flights, respectively, by calculating the link travel times on a node-link model of the airport surface movement. The prediction performance of the proposed method was evaluated using recorded track data from Incheon International Airport during April 2015. Furthermore, the prediction performance was compared with three other methods well known for calculating unimpeded taxi times. The root-mean-squared error of the unimpeded taxi times predicted by the proposed method showed a prediction error of about 1 min for the departure and arrival flights. Moreover, the results showed significant improvement in the predictions of the unimpeded taxi time compared to the other methods.