Teleseismic P and SH body waves are used in a finite-fault, waveform inversion for the rupture history of the 23 November 1977 western Argentina earthquake. This double event consists of a smaller foreshock (M0 = 5.3 ?? 1026 dyn-cm) followed about 20 s later by a larger main shock (M0 = 1.5 ?? 1027 dyn-cm). Our analysis indicates that these two events occurred on different fault segments: with the foreshock having a strike, dip, and average rake of 345??, 45??E, and 50??, and the main shock 10??, 45??E, and 80??, respectively. The foreshock initiated at a depth of 17 km and propagated updip and to the north. The main shock initiated at the southern end of the foreshock zone at a depth of 25 to 30 km, and propagated updip and unilaterally to the south. The north-south separation of the centroids of the moment release for the foreshock and main shock is about 60 km. The apparent triggering of the main shock by the foreshock is similar to other earthquakes that have involved the failure of multiple fault segments, such as the 1992 Landers, California, earthquake. Such occurrences argue against the use of individual, mapped, surface fault or fault-segment lengths in the determination of the size and frequency of future earthquakes.