The largest events in the 1980 Mammoth Lakes earthquake sequence show a discrepancy between fault mechanisms which are determined on the basis of the local short-period first motions and those determined by modeling of long-period regional and teleseismic waveforms. The short-period solutions are left-lateral strike-slip on north-striking, near vertical planes. The long-periods invariably require a much more moderately dipping fault plane with a significant dip-slip (normal) component. Persistence of disagreements between short- and long-period polarities to teleseismic distances suggests that the source-time functions are complicated and may be responsible for at least part of the discrepancy. In addition, there seems to be a systematic difference between local short-period polarities and teleseismic long-period polarities that is related to travel paths across portions of Long Valley Caldera. It is possible that a low velocity zone related to recent magmatic activity is causing the deflection of local seismic rays, thus distorting the fault plane projection.