The November 20, 1960 Peru earthquake excited a tsunami that was anomalously large for an earthquake of moderate magnitude (M 6.75 PAS). Body wave first motions combined with observed Love to Rayleigh wave spectral ratios indicate slip along a shallowly dipping thrust fault rather than a single force mechanism suggestive of a landslide. Deconvolution of SH waveforms shows a time function consisting of 4 subevents with a total duration of 110 s. Using this time function, body waveform inversion yields a very shallowly dipping thrust faulting mechanism (strike 0¿ , dip: 6¿, slip: 100¿) and a seismic moment of 1.9¿1027 dyn cm. However, the time function obtained from the band limited body waveforms cannot fit the surface wave data which show a large increase in amplitude with period and a moment of 3.4¿1027 dyn cm. Both types of data can be modeled by combining the body wave time function and a 130 s cosine time function representing a longer period component. This event represents the largest earthquake to occur along this relatively aseismic portion of the Peruvian coast. The tsunami excitation is not anomalous relative to the measured seismic moment and moment magnitude (Mw 7.6). The disparity between tsunami height and surface wave magnitude results from underestimation of the earthquake size by conventional magnitude scales due to the long source duration. ¿ American Geophysical Union 1990