A multilevel multi-modal routing navigating method relates to a routing navigation, which is implemented by establishing a road central line node table and an arc section table, generating topology texts T1 and T2, inputting starting point S and destination D, reading road central line and traffic lane attribute data, and calculating optimal routes R1 and R2 based on T1 and T2; establishing actual traffic lane topological structure T3, and recording physical traffic lane ID, virtual traffic lane ID as well as associated starting point ID and destination ID; reading actual traffic lane attribute data, and calculating optimal route R3 based on T3; according to R1, generating a buffer area B1 with certain range d1 in a dynamic way, selecting out architecture set L1 in B1, expressing L1 in a 2D visible window, selecting out landmark architecture set L2 in L1, expressing L2 in the 2.5D visible window, according to the coordinate position of a driver, dynamically calling a buffer area B2 of d2, selecting real time visible landmark L3 containing L2 in B2, and dynamically calling in road traffic relevant facilities in B2.