A new, fully automated processing algorithm based on network beamforming of raw automatic detection data has been implemented to improve the analyst’s capability to identify the seismic depth phases pP and sP, and thereby to improve focal-depth estimation accuracy in seismic bulletins. This algorithm has been tested on data recorded from a sample of 150 subcrustal earthquakes located in the Pamir-Hindu Kush, Hokkaido, and central Honshu seismic zones and has successfully identified candidate depth phases for over 70% of the events analyzed, including some with magnitudes as low as mb 3.75. Several supplemental analyst tools for validating such candidate depth phases identified by the network-beamforming algorithm have also been implemented and evaluated in a preliminary fashion. These include the F detector (Blandford, 1974), which provides a quantitative statistical measure of whether a particular detection has a vector slowness consistent with that expected for a depth phase, and the Pearce (1977, 1980) focal-mechanism algorithm, which provides a means for determining whether the observed candidate depth-phase amplitude patterns across the network of observing stations are consistent with the regional tectonic environments in which the earthquakes occur. Results of the tests conducted to date indicate that the use of these automated analyst tools should significantly improve the confident identification of pP and sP phases for inclusion in hypocenter inversion analyses.