Initial focal mechanism determinations for the 29 November 1975 Kalapana, Hawaii, earthquake indicated discrepancy between the mechanism determined from teleseismic data by Ando and the mechanism determined using data from the local U.S. Geological Survey network surrounding the epicenter region. The resolution of this difference is crucial to correctly understand this earthquake, as well as to understand the tectonics of the south flank of Kilauea volcano. When a model with a low-velocity layer at the base of the crust is used for projection back to the focal sphere for the local network mechanisms, the discrepancy vanishes. To further investigate this result, focal mechanisms were determined using several contrasting models for a set of well-recorded earthquakes. A large number of these earthquakes have mechanisms identical to the main shock when the low-velocity layer model is used. Dispersion of P and T axes is also minimized by use of this model. A low-angle slip direction, favored for the main shock and typical of most other solutions, exhibits remarkable stability normal to the east rift zone of Kilauea. Our results suggest a tectonic model, similar in nature to that proposed by Ando, in which the south flank of Kilauea consists of a mobile block of crust which is relatively free to move laterally on a low-strength zone at about 10 km depth. Forceful injection of magma along the rift zones provides the loading stress which is released by catastrophic failure in the weak, horizontal layer in a cycle of perhaps 100 yr.