With the rapid growing popularity of cellular phones, cellular vehicle probe (CVP)-based travel time information systems become comparatively advantageous. Due to the vague positioning problem of CVP, to directly determine the travel path with the closest travel time may lead to an erroneous result. This paper proposes a virtual trip line matching model to determine the travel path of a mobile user by using the Yen's K-shortest paths algorithm to obtain potential paths and the latent class model (LCM) to develop the map-matching model. A simplified network is used to estimate the LCM models and to validate the performance of the proposed model under various traffic conditions simulated by Paramics traffic simulator. The results show that prediction accuracy of travel paths can be improved by approximate 45%. Accordingly, the proposed model remarkably outperforms than the traditional model.