A theory presented in an earlier paper (Mooney, 1974) has been used to compute seismic waveforms produced by a surface impact. Surface detectors are assumed which measure displacement, velocity, acceleration, or strain in either vertical or horizontal components. A physically plausible source waveform is considered which is unidirectional, symmetrical about a peak, and with zero slopes at beginning and end. Theoretical waveforms are presented for models simulating projectile, hammer, and weight-drop impacts onto granite, concrete, and soil.The waveforms show small but clear P arrivals, large Rayleigh pulses, and no recognizable S waves. P waves are larger on the horizontal sensors, Rayleigh on the vertical. The waveforms differ markedly from one detector type to another and show little resemblance to the source waveform.Distinctive pulse length and amplitude pulse parameters are selected. Pulse lengths are shown to be relatively insensitive to detector distance, hence conversion factors can be obtained from which to infer source pulse length. These factors were applied to accelerometer waveforms obtained experimentally from steel pellet impacts onto granite, yielding a source pulse length of 30 mu sec. This value was used to compute theoretical waveforms which show satisfactory agreement with the experimental waveforms.The measured amplitude parameters are shown to depend upon detector distance r and source pulse length T and amplitude H asEquationwhere n = 1, 1/2 for P and Rayleigh waves, respectively; p = 0, 1/2 for P and Rayleigh waves; and m = 0,1,2,1 for detectors which measure displacement, particle velocity, acceleration, and horizontal strain.An increase in Poisson's ratio v for the medium produces increased time separation between P and Rayleigh arrivals, an increase in P pulse length and amplitude, no changes in Rayleigh pulse length, and a decrease in Rayleigh amplitude as v (super -1) 2/.