Relative locations of 323 large aftershocks (M 3.0 or greater) in the period from 5 October 1985 to 25 March 1988 show that the Ms 6.6 event on 5 October 1985 initiated at 62.208°N, 124.217°W, about 2.5 km northeast of the Ms 6.9 main shock on 23 December 1985. The overall aftershock distribution suggests the October rupture was primarily a west-dipping, low-angle thrust. In subsequent aftershock activity, the main rupture plane was marked by a distinct quiescent area of about 200 km2 that persisted until the 23 December event. Most of the stress drop and slip occurred in this area. Following the 23 December rupture, a similar sized quiescent zone was also observed; however, it was only evident during the first 24 hr of the aftershock sequence, and the area was about 50 per cent too small to yield the overall stress drop. The additional area appeared to come from secondary rupture zones that developed coincident with the main shock rupture. Precise locations of 182 small (M 3.0 or less) aftershocks recorded during a third field survey from 12 to 21 September 1986 indicated at least one and probably three high-angle faults. Composite mechanism solutions showed thrust faulting except in a region directly south of the main shock rupture areas where there is a bend in one of the secondary fault zones and a concentration of aftershock activity. Mechanism solutions calculated for five of the largest aftershocks in the same region also indicated a similar variability. Development of secondary fault zones explained the increased complexity of the December event and may also provide an explanation for the vertical peak acceleration exceeding 2 g that was recorded about 10 sec after the December rupture initiated.