The invention relates to a traffic big data-based travel time extraction, prediction and query method. The online processing part of the method includes the following steps that: S1, historical floating vehicle data are collected, a historical and real-time database is constructed; S2, the query modes of users are demarcated; S3, timing is carried out automatically, and floating vehicle data are collected in real time; S4, whether current time points t are equal to termination time points of the query of the users is judged; S5, real-time travel time fitting analysis is carried out on other road sections or paths in a road network which are not processed; S6, whether all the road sections or paths are traversed is judged; and S7, whether the query is terminated is judged. The method has the advantages of high precision of road network travel time extraction, high data processing speed and high stability. The method according to which a large number of floating vehicle data are utilized to acquire and predict road network travel time, can carry out real-time, predictive and historical road network travel time processing, has higher real-time performance of data storage, reading and writing and high positioning accuracy, and can meet the needs of the users and enable great user experience.