Sliding-window cross-correlation method is firstly adopted to identify sPn phase, and to constrain focal depth from regional seismograms, by measuring the time separation between sPn and Pn phases. We present the focal depths of the 17 moderate-sized aftershocks (M S ≥5.0) of the Wenchuan M S 8.0 earthquake, using the data recorded by the regional seismic broadband networks of Shaanxi, Qinghai, Gansu, Yunnan and Sichuan. Our results show focal depths of aftershocks range from 8 to 20 km, and tend to cluster at two average depths, separate at 32.5°N, i.e., 11 km to the south and 17 km to the north, indicating that these aftershocks are origin of upper-to-middle crust. Combined with other results, we suggest that the Longmenshan fault is not a through-going crustal fault and the Pingwu-Qingchuan fault may be not the northward extension of the Longmenshan thrust fault.Wenchuan M S 8.0 earthquake, aftershocks, sPn phaseThe Wenchuan M S 8.0 earthquake is the largest intraplate earthquake in china in the latest 30 years, after the M S 8.1 Kunlunshan earthquake. Up to now, more than ten thousands of aftershocks were triggered and still continuously occurred. Among them, more than 39 aftershocks with magnitude greater than 5.0 occurred along the Longmenshan thrust belt, which provide us an opportunity to delineate the depth extension of the Longmenshan fault, and to address the problems whether the Longmenshan fault is a crustal extend fault suggested by deep seismic sounding data [1,2] , and the Pingwu-Qingchuan fault is the northern extension of the Longmenshan fault. Focal depth is one of the critical earthquake source parameters. Distribution of focal depth is essential for understanding stress regime at the depth where earthquake was located, regional seismicity and seismic hazards [3][4][5][6] ; it plays a particularly important role in relating seismicity to geologic structure and delineating active faults [7] . Reliably determined focal depth is also useful for establishing accurate epicentral location and origin time [8] .However, in source parameters, focal depth is difficult to be determined and sometimes not well constrained. In the routine location determination, focal depth is mainly constrained by modeling or inversion of teleseismic first arrival of P wave. Since there is a strong trade-off between origin time and depth, especially for shallow-focus earthquake, error in depth determination is serious when local or regional stations nearby the source are unavailable. When the P-wave inversion is not convergent in focal depth, the time separation between depth phase sP (or pP) and its reference phase P is used to constrain focal depth by teleseismic seismograms modeling [3,[6][7][8][9][10] . Depth analysis is usually based on a generic global model, which does not consider the region-specific upper crust, and due to poor azimuthal coverage of seismic stations, thus the error in focal depth is significant even for large events (M w >5) [11] . Many institutions often use a fixed value for those events with fo...