Cross‐correlating the output of two widely separated receivers may fail to provide a means to estimate the arrival time difference of the first arriving signals when the cross‐correlation of accompanying multipaths have amplitudes that may be confused with that between the first arrivals. Based on information in two auto‐correlation functions, a new, robust, and efficient method is described for identifying which peak in the cross‐correlation function corresponds to the arrival time difference between the first arrivals. The information in auto‐correlation functions is often sufficient to identify the relative arrival times of all the multipaths at each receiver. The technique relies on (1) a high signal‐to‐noise ratio following auto‐ and cross‐correlation, and (2) time series which have been preprocessed to suppress all signals except those emanating from a single source. The technique works even when (1) the number of multipaths is unknown, (2) the received signal is not visible above the noise, (3) the transmission time is unknown, (4) the transmitted waveform is unknown, (5) the spatial coordinates of the multipaths are unknown or are impractical to estimate, and (6) the differential Doppler between multipaths is not too large. These techniques may be useful for localizing transient signals and mapping the environment with tomography.