The conventional ray-based method for the determination of passive seismic events is performed by minimizing traveltime differences between pairs of events or stations, which needs to identify and pick arrival times of the events in the seismogram. This method usually suffers from strong heterogeneity and background noise, which may make it less sensitive to source localization. Time-reversal method is another technique for identifying and locating passive seismic events, which is based on concentrating the back-propagated wave energy from many receivers at a specific time and location inside a medium. It is more robust and effective in heterogeneous media. However, it is not well suited to strong background noise and limited acquisition aperture. To go beyond the limitations of these two methods, we present a solution by constructing maximum-amplitude paths of the back-propagated wavefields instead of rays, and estimating the excitation time and location for passive seismic events by minimizing the normalized overall distance of all paths. The numerical case studies for simple and complex models illustrate a good performance of the proposed method.