We apply the wavelet transform to seismic signals for the purpose of automatically identifying the P and S phase arrivals of seismic events. In this article, an algorithm is presented that locates these arrivals in single-station three-component short-period seismograms using polarization and amplitude information contained in the wavelet transform coefficients of the signals. The main idea is that strong features of the seismic signal appear in the wavelet coefficients across several scales. The first step in the algorithm is the wavelet decomposition of each component of a three-component short-period seismogram. The resulting multi-scalar representation is used to construct “locator” functions that identify the P and S arrivals. The P locator function is constructed by using polarization information across scales, and the S locator function is constructed using transverse over radial amplitude information across scales. These functions prove to be very effective at identifying the important P and S arrivals in the test data. The results are compared with arrival times picked by an analyst.