The MS 7.2 Grand Banks earthquake of 1929 was one of the largest and the most fatal earthquakes to have occurred in Canada, with most of the death and destruction having been caused by a tsunami and submarine landslide associated with the earthquake. It has been suggested (Hasegawa and Kanamori, 1987) that a single-force (landslide) mechanism was more consistent with the data than was a double-couple source and that therefore the event was not an earthquake. However, that particular study considered only four double-couple solutions and left many unanswered questions, in particular with respect to the source time function and sediment volume involved. Here, a larger number of seismograms are used to examine the full range of double-couple solutions to determine more definitively the nature of the event. Waveform modeling using both forward and inverse methods indicates that this event was an earthquake, with a complex source mechanism. The first and largest subevent was a strike-slip double-couple event occurring on a northwest-striking plane. Two later subevents were probably strike-slip double couples on northeast-striking planes, but other mechanisms cannot be completely ruled out. The first subevent has a well-constrained focal depth of 20 ± 2 km. The second and third subevents also appear to have occurred at 20 km, but are constrained only to within ±5 km. These depths provide further evidence that the event was not a landslide. The sum of the subevent moments corresponds to an Mw of 7.2 ± 0.3, which is in close agreement with the Mw of 7.1 ± 0.1 obtained by the CMT method using long-period data. These Mw's are also consistent with the Ms of 7.2 (±0.3) and mB of 7.1 (±0.2) calculated directly from the seismograms. Modeling of the seafloor displacement for this mechanism indicates that the tsunami was generated by the landslide and not directly by the earthquake.