In large-scale regional evacuation, public pre-evacuation time estimation is an important link of evacuation analysis and emergency management. This paper redefines the concept of "pre-evacuation time" based on the characteristics of unconventional emergencies. A new prediction model of traffic loading rate for large-scale evacuations based on a real case survey was proposed. The model takes into account two groups of people who have received evacuation order/message (EOM) and have not received EOM. The analysis shows that the probability distributions of the pre-evacuation times of those who received and did not receive an EOM conform to Weibull distributions. The earlier the EOM is received, the shorter the average pre-evacuation time of the people is. At the same time, to further study the characteristics of the pre-evacuation time of the evacuating population, the distribution of the pre-evacuation time is discussed from three aspects: the time of receipt of the EOM, the source of EOM, and the channels of receipt of the EOM. The new model and method can more accurately estimate the loading rate of evacuation traffic flow in large-scale evacuation when the public cannot be accurately informed of the EOM time. (C) 2020 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.